AE-WGAN-GP.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyNy51yhH6faQOZtzuhvjvFC"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# Environment setup\n",
        "\n",
        "## Import necessary packages and functions"
      ],
      "metadata": {
        "id": "ElPEP8AqxbTZ"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 27,
      "metadata": {
        "id": "_Hl9i9wxA1WB"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import pickle\n",
        "import random\n",
        "import time\n",
        "import pprint as pp\n",
        "import math\n",
        "\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.model_selection import train_test_split\n",
        "import tensorflow as tf"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='5'>Define parameters</font>"
      ],
      "metadata": {
        "id": "qTik8n2IPZPO"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "GIT_USERNAME = \"dado330\"\n",
        "GIT_MAIL = \"messina.davide.statistician@gmail.com\"\n",
        "GIT_REPO = \"https://github.com/dado330/AE-WGAN-GP\"\n",
        "\n",
        "train_AE = True\n",
        "train_WGAN_GP = False\n",
        "train_from_partial = False\n",
        "\n",
        "batchsize = 20\n",
        "Z_DIM = 512\n",
        "random.seed(123)"
      ],
      "metadata": {
        "id": "x8PmxJ3mPZbM"
      },
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Setup git and repo\n",
        "\n",
        "### Git parameter to be run only once"
      ],
      "metadata": {
        "id": "PjHebxjn8E33"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!git clone $GIT_REPO"
      ],
      "metadata": {
        "id": "yU9JBz1h8FC-",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "79cb906f-74db-41e6-be1d-f3dc8e77364d"
      },
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "fatal: destination path 'AE-WGAN-GP' already exists and is not an empty directory.\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='4'>Pull changes before running the script</font>"
      ],
      "metadata": {
        "id": "VKglv2hhSaAB"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "%%bash\n",
        "cd /content/AE-WGAN-GP\n",
        "git pull origin main"
      ],
      "metadata": {
        "id": "5X1YaAvwSZtj",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "46e981bd-a9d4-45a0-90f4-d06dd0318a0e"
      },
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Already up to date.\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "From https://github.com/dado330/AE-WGAN-GP\n",
            " * branch            main       -> FETCH_HEAD\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Check if GPU has been loaded correctly"
      ],
      "metadata": {
        "id": "OMivdUQGzGvt"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "if train_AE or train_WGAN_GP:\n",
        "    device_name = tf.test.gpu_device_name()\n",
        "    if device_name != '/device:GPU:0':\n",
        "        raise SystemError('GPU device not found')\n",
        "    print('Found GPU at: {}'.format(device_name))"
      ],
      "metadata": {
        "id": "TBsBrGaEClo6"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "# AUTOENCODER (AE)\n",
        "\n",
        "## Encoder definition\n",
        "\n",
        "Input is with dimension 18 and has four dense layers of increasing size 64, 128, 256, 512\n",
        "\n",
        "Internal layers have Selu as activation function while the output layer has the sigmoid function\n",
        "\n",
        "The model is fairly big with respect to the data so I used l1 as kernel regularizer to make it possible for the model to set some of the nodes to 0"
      ],
      "metadata": {
        "id": "9b9ai37MzO6d"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "class Encoder(tf.keras.Model):\n",
        "    def __init__(self):\n",
        "        super(Encoder, self).__init__()\n",
        "        self.Encoder_DIMS = [64, 128, 256, 512]\n",
        "        self.dense_layers = [tf.keras.layers.Dense(dim, activation=tf.nn.leaky_relu, kernel_regularizer='l1')\n",
        "                             for dim in self.Encoder_DIMS[:-1]]\n",
        "        self.output_layer = tf.keras.layers.Dense(self.Encoder_DIMS[-1], activation=tf.nn.tanh)\n",
        "\n",
        "    def call(self, x):\n",
        "        for s in range(len(self.dense_layers)):\n",
        "            x = self.dense_layers[s](x)\n",
        "        x = self.output_layer(x)\n",
        "        return x"
      ],
      "metadata": {
        "id": "lIuFRwHXxZkf"
      },
      "execution_count": 44,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='5'>Decoder definition</font>\n",
        "\n",
        "Input is with dimension 512 and has four dense layers of decreasing size 256, 128, 64, 18\n",
        "\n",
        "Internal layers have Selu as activation function while the output layer has the sigmoid function. The sigmoid in this is necessary since each cell of our dataset is binary\n",
        "\n",
        "The model is fairly big with respect to the data so I used l1 as kernel regularizer to make it possible for the model to set some of the nodes to 0"
      ],
      "metadata": {
        "id": "LtY3jdijzpSq"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "class Decoder(tf.keras.Model):\n",
        "    def __init__(self):\n",
        "        super(Decoder, self).__init__()\n",
        "        self.Decoder_DIMS = [256, 128, 64, 18]\n",
        "        self.dense_layers = [tf.keras.layers.Dense(dim, activation=tf.nn.leaky_relu, kernel_regularizer='l1')\n",
        "                             for dim in self.Decoder_DIMS[:-1]]\n",
        "        self.output_layer = tf.keras.layers.Dense(self.Decoder_DIMS[-1], activation=tf.nn.sigmoid)\n",
        "\n",
        "    def call(self, x):\n",
        "        for j in range(len(self.dense_layers)):\n",
        "            x = self.dense_layers[j](x)\n",
        "        x = self.output_layer(x)\n",
        "        return x"
      ],
      "metadata": {
        "id": "UAp9WeepzNOc"
      },
      "execution_count": 45,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Additional parameters\n",
        "\n",
        "Initialize the model for the autoencoder and create the checkpoint manager"
      ],
      "metadata": {
        "id": "IeJq2nFhizYw"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "encoder = Encoder()\n",
        "decoder = Decoder()\n",
        "\n",
        "checkpoint_directory_ae = \"/content/AE-WGAN-GP/AE_checkpoints\"\n",
        "\n",
        "checkpoint_ae = tf.train.Checkpoint(encoder=encoder, decoder=decoder)\n",
        "manager_ae = tf.train.CheckpointManager(checkpoint_ae, directory=checkpoint_directory_ae, max_to_keep=5)"
      ],
      "metadata": {
        "id": "KCRd8EU4z9kk"
      },
      "execution_count": 46,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='4'>Optimizer and main loss</font>\n",
        "\n",
        "Define the optimizer as Adam with relative low learning rate and betas. Added weigth decay to improve regularization (similar to l2)\n",
        "\n",
        "Use bynary crossentropy as the main loss function"
      ],
      "metadata": {
        "id": "WMP3pLqRz9af"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "ae_optimizer = tf.keras.optimizers.AdamW(learning_rate=0.0001, weight_decay=0.0004,\n",
        "                                             beta_1=0.7, beta_2=0.99)\n",
        "\n",
        "bce = tf.keras.losses.BinaryCrossentropy(from_logits=False)\n"
      ],
      "metadata": {
        "id": "ut6iDaaMi0Ls"
      },
      "execution_count": 47,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='4'>Learning rate scheduler</font>\n",
        "\n",
        "Define a custom learning rate scheduler to decrease and fine tune the model after the first epochs"
      ],
      "metadata": {
        "id": "-Dyi77pP0lpV"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def set_learning_rate(epoch):\n",
        "    if epoch > 20:\n",
        "        ae_optimizer.learning_rate = 1e-5\n",
        "    elif epoch > 10:\n",
        "        ae_optimizer.learning_rate = 3e-5\n",
        "    elif epoch > 3:\n",
        "        ae_optimizer.learning_rate = 6e-5"
      ],
      "metadata": {
        "id": "PnE47D_i0mF3"
      },
      "execution_count": 48,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='5'>Training step</font>\n",
        "\n",
        "The loss function is not the vanilla binary crossentropy because otherwise the model was collapsing to all 0 or mode collapsing"
      ],
      "metadata": {
        "id": "sZ6w9p7H11vR"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def ae_step(real):\n",
        "    with (tf.GradientTape() as disc_tape):\n",
        "        synthetic = decoder(encoder(real))\n",
        "\n",
        "        main_sparsity_penalty = - tf.math.log(tf.reduce_mean(synthetic + 1e-10)) + tf.math.log(\n",
        "            tf.reduce_mean(tf.cast(real, tf.float32) + 1e-10))\n",
        "        sparsity_penalty = - tf.math.log(1 - main_sparsity_penalty)\n",
        "        autoenc_loss = tf.cast(bce(synthetic, real), tf.float32) + main_sparsity_penalty\n",
        "        if np.random.uniform(size=1) > 0.99:\n",
        "            print(\"synthetic\")\n",
        "            print(tf.round(synthetic).numpy())\n",
        "            print(\"real\")\n",
        "            print(real.numpy())\n",
        "\n",
        "    gradients_of_autoencoder = disc_tape.gradient(autoenc_loss,\n",
        "                                                  encoder.trainable_variables + decoder.trainable_variables)\n",
        "    ae_optimizer.apply_gradients(zip(gradients_of_autoencoder,\n",
        "                                     encoder.trainable_variables + decoder.trainable_variables))\n",
        "    return autoenc_loss"
      ],
      "metadata": {
        "id": "XCzIjlMs11jN"
      },
      "execution_count": 49,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Train the model\n",
        "\n",
        "Load the data, divide it in train and test datasets"
      ],
      "metadata": {
        "id": "Z-vZ9qICiWSs"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "data = np.load(\"/content/AE-WGAN-GP/Input_data/D3_events_ALL_OUTCOMES_ML.npy\", allow_pickle=True)\n",
        "train_data, test_data = train_test_split(data, test_size=0.1)\n",
        "dataset_train = tf.data.Dataset.from_tensor_slices(train_data).shuffle(\n",
        "        1000, reshuffle_each_iteration=True).batch(batchsize, drop_remainder=True)"
      ],
      "metadata": {
        "id": "GQgZEwsbiWjd"
      },
      "execution_count": 41,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "if train_AE:\n",
        "    epochs = 50\n",
        "    steps = len(dataset_train)\n",
        "\n",
        "    for epoch in range(epochs):\n",
        "        start_time = time.time()\n",
        "        set_learning_rate(epoch)\n",
        "        aeloss = 0.0\n",
        "        for batch_sample in dataset_train:\n",
        "            aeloss += ae_step(batch_sample)\n",
        "        duration_epoch_ae = time.time() - start_time\n",
        "        format_str = 'epoch: %d, aeloss = %.3f (%.2f)'\n",
        "        print(format_str % (epoch, aeloss / steps / batchsize * 10000, duration_epoch_ae))\n",
        "\n",
        "else:\n",
        "    checkpoint_ae.restore(manager_ae.latest_checkpoint).expect_partial().assert_existing_objects_matched()"
      ],
      "metadata": {
        "id": "1psXk0NZzciX",
        "outputId": "9ab17b64-6bdd-40f7-c09e-21488e71d0d6",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "epoch: 0, aeloss = 2639.727 (2.54)\n",
            "epoch: 1, aeloss = 1256.627 (2.54)\n",
            "epoch: 2, aeloss = 681.442 (2.63)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 3, aeloss = 666.782 (2.54)\n",
            "epoch: 4, aeloss = 664.938 (2.25)\n",
            "epoch: 5, aeloss = 652.690 (2.18)\n",
            "epoch: 6, aeloss = 656.853 (2.54)\n",
            "epoch: 7, aeloss = 642.859 (2.66)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 8, aeloss = 628.229 (2.21)\n",
            "epoch: 9, aeloss = 618.826 (2.20)\n",
            "epoch: 10, aeloss = 614.237 (2.21)\n",
            "epoch: 11, aeloss = 608.781 (2.54)\n",
            "epoch: 12, aeloss = 601.545 (2.76)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]\n",
            "epoch: 13, aeloss = 585.542 (2.16)\n",
            "epoch: 14, aeloss = 573.257 (2.25)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 15, aeloss = 549.621 (2.21)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "epoch: 16, aeloss = 524.255 (2.54)\n",
            "synthetic\n",
            "[[0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "synthetic\n",
            "[[0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 17, aeloss = 498.591 (5.10)\n",
            "epoch: 18, aeloss = 482.755 (2.81)\n",
            "epoch: 19, aeloss = 465.402 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]]\n",
            "epoch: 20, aeloss = 453.768 (2.54)\n",
            "epoch: 21, aeloss = 439.337 (2.69)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 22, aeloss = 441.358 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 23, aeloss = 436.227 (2.96)\n",
            "epoch: 24, aeloss = 433.198 (2.19)\n",
            "epoch: 25, aeloss = 435.035 (5.10)\n",
            "epoch: 26, aeloss = 428.938 (2.20)\n",
            "epoch: 27, aeloss = 427.810 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]\n",
            "epoch: 28, aeloss = 426.467 (2.55)\n",
            "epoch: 29, aeloss = 420.008 (5.10)\n",
            "epoch: 30, aeloss = 423.928 (2.54)\n",
            "epoch: 31, aeloss = 422.285 (2.54)\n",
            "epoch: 32, aeloss = 417.519 (2.24)\n",
            "epoch: 33, aeloss = 417.305 (2.94)\n",
            "epoch: 34, aeloss = 420.225 (2.54)\n",
            "epoch: 35, aeloss = 415.434 (2.54)\n",
            "epoch: 36, aeloss = 409.357 (2.54)\n",
            "epoch: 37, aeloss = 413.485 (2.54)\n",
            "epoch: 38, aeloss = 415.025 (2.82)\n",
            "epoch: 39, aeloss = 407.370 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 40, aeloss = 408.355 (2.18)\n",
            "epoch: 41, aeloss = 415.703 (2.54)\n",
            "epoch: 42, aeloss = 414.444 (2.45)\n",
            "epoch: 43, aeloss = 409.379 (5.10)\n",
            "epoch: 44, aeloss = 413.109 (2.21)\n",
            "epoch: 45, aeloss = 410.729 (2.21)\n",
            "synthetic\n",
            "[[0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]\n",
            "real\n",
            "[[0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]\n",
            "epoch: 46, aeloss = 412.988 (2.23)\n",
            "epoch: 47, aeloss = 409.550 (5.10)\n",
            "epoch: 48, aeloss = 409.662 (2.54)\n",
            "epoch: 49, aeloss = 405.683 (2.54)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Train the autoencoder and save it"
      ],
      "metadata": {
        "id": "PkMtO0qNiRPy"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "if train_AE:\n",
        "    epochs = 50\n",
        "    steps = len(dataset_train)\n",
        "\n",
        "    for epoch in range(epochs):\n",
        "        start_time = time.time()\n",
        "        set_learning_rate(epoch)\n",
        "        aeloss = 0.0\n",
        "        for batch_sample in dataset_train:\n",
        "            aeloss += ae_step(batch_sample)\n",
        "        duration_epoch_ae = time.time() - start_time\n",
        "        format_str = 'epoch: %d, aeloss = %.3f (%.2f)'\n",
        "        print(format_str % (epoch, aeloss / steps / batchsize * 10000, duration_epoch_ae))\n",
        "\n",
        "    manager_ae.save()\n",
        "else:\n",
        "    checkpoint_ae.restore(manager_ae.latest_checkpoint).expect_partial().assert_existing_objects_matched()\n"
      ],
      "metadata": {
        "id": "ncxmY2KCDh8c",
        "outputId": "c8608dce-3440-4c93-b3bf-0501a5d42c86",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "epoch: 0, aeloss = 1559.497 (5.14)\n",
            "epoch: 1, aeloss = 476.697 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 1. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 2, aeloss = 238.158 (2.21)\n",
            "epoch: 3, aeloss = 137.087 (2.54)\n",
            "epoch: 4, aeloss = 89.965 (2.92)\n",
            "epoch: 5, aeloss = 70.292 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 6, aeloss = 53.287 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "epoch: 7, aeloss = 40.454 (2.54)\n",
            "epoch: 8, aeloss = 30.519 (2.54)\n",
            "epoch: 9, aeloss = 24.266 (2.77)\n",
            "synthetic\n",
            "[[0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 10, aeloss = 18.108 (2.15)\n",
            "epoch: 11, aeloss = 15.043 (2.20)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 12, aeloss = 13.507 (2.23)\n",
            "epoch: 13, aeloss = 11.278 (2.17)\n",
            "synthetic\n",
            "[[1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 14, aeloss = 10.111 (5.11)\n",
            "epoch: 15, aeloss = 8.845 (2.54)\n",
            "epoch: 16, aeloss = 7.753 (2.19)\n",
            "epoch: 17, aeloss = 6.697 (10.22)\n",
            "epoch: 18, aeloss = 5.380 (2.54)\n",
            "epoch: 19, aeloss = 4.918 (2.54)\n",
            "epoch: 20, aeloss = 4.053 (2.94)\n",
            "synthetic\n",
            "[[1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "epoch: 21, aeloss = 3.731 (2.54)\n",
            "epoch: 22, aeloss = 3.422 (2.54)\n",
            "epoch: 23, aeloss = 3.034 (2.54)\n",
            "epoch: 24, aeloss = 3.149 (2.28)\n",
            "epoch: 25, aeloss = 2.872 (2.78)\n",
            "epoch: 26, aeloss = 2.551 (2.54)\n",
            "epoch: 27, aeloss = 2.179 (2.24)\n",
            "epoch: 28, aeloss = 2.350 (5.10)\n",
            "epoch: 29, aeloss = 1.977 (2.64)\n",
            "epoch: 30, aeloss = 2.011 (2.23)\n",
            "synthetic\n",
            "[[1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 31, aeloss = 1.872 (2.54)\n",
            "epoch: 32, aeloss = 1.657 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]]\n",
            "epoch: 33, aeloss = 1.407 (2.50)\n",
            "epoch: 34, aeloss = 1.421 (5.10)\n",
            "synthetic\n",
            "[[0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 35, aeloss = 1.206 (2.54)\n",
            "epoch: 36, aeloss = 0.973 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]]\n",
            "epoch: 37, aeloss = 1.055 (2.69)\n",
            "epoch: 38, aeloss = 0.920 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]]\n",
            "epoch: 39, aeloss = 0.878 (2.54)\n",
            "epoch: 40, aeloss = 0.863 (2.54)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 41, aeloss = 0.774 (2.54)\n",
            "epoch: 42, aeloss = 0.849 (5.11)\n",
            "epoch: 43, aeloss = 0.724 (2.18)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 44, aeloss = 0.561 (2.25)\n",
            "epoch: 45, aeloss = 0.457 (2.21)\n",
            "epoch: 46, aeloss = 0.399 (5.10)\n",
            "synthetic\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "real\n",
            "[[0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 0. 0. 1. 0. 0. 0. 0. 0. 0. 0. 0. 0.]]\n",
            "epoch: 47, aeloss = 0.413 (2.19)\n",
            "epoch: 48, aeloss = 0.461 (2.16)\n",
            "epoch: 49, aeloss = 0.310 (2.54)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Results\n",
        "\n",
        "Print the loss of the autoencoder for both train and test data.\n",
        "\n",
        "The model reconstruct very well the original data"
      ],
      "metadata": {
        "id": "QC4-tVGFridN"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "aeloss_train = ae_step(train_data)\n",
        "aeloss_test = ae_step(test_data)\n",
        "format_str = 'train_aeloss: %.3f, test_aeloss = %.3f'\n",
        "print(format_str % (aeloss_train / len(train_data) * 10000, aeloss_test / len(test_data) * 10000))"
      ],
      "metadata": {
        "id": "9trDPqSfriAW",
        "outputId": "d4f1a20a-025f-45cf-ab12-e75b163fbe05",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": 43,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "train_aeloss: 0.014, test_aeloss = -0.178\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP)\n",
        "\n",
        "## Generator definition\n",
        "\n",
        "Input is with dimension 512 and has three dense layers of fixed size 512\n",
        "\n",
        "Size the layers is fixed beacuse it is needed for the passtrougth\n",
        "\n",
        "Internal layers have Selu as activation function while the output layer has the sigmoid function. The sigmoid in this is necessary since it should be the same of the autoencoder.\n",
        "\n",
        "Added batch normalization before every dense layer and shortcut\n",
        "connections (as MedGAN)"
      ],
      "metadata": {
        "id": "2AAJmkBUz6tP"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# class Generator(tf.keras.Model):\n",
        "#     def __init__(self):\n",
        "#         super(Generator, self).__init__()\n",
        "#         self.G_DIMS = [512, 512, 512]\n",
        "#         self.batch_norm_layers = [tf.keras.layers.BatchNormalization(epsilon=1e-5) for _ in self.G_DIMS[:-1]]\n",
        "#         self.dense_layers = [tf.keras.layers.Dense(dim, activation=tf.nn.selu, kernel_regularizer='l1')\n",
        "#                              for dim in self.G_DIMS[1:-1]]\n",
        "#         self.output_layer = tf.keras.layers.Dense(self.G_DIMS[-1], activation=tf.nn.sigmoid)\n",
        "\n",
        "#     def call(self, x, training):\n",
        "#         for k in range(0, len(self.G_DIMS[:-2])):\n",
        "#             x1 = self.dense_layers[k](self.batch_norm_layers[k](x, training=training))\n",
        "#             x += tf.keras.layers.Add()([x1, x])\n",
        "#         x2 = self.output_layer(self.batch_norm_layers[-1](x, training=training))\n",
        "#         x += tf.keras.layers.Add()([x2, x])\n",
        "#         return x\n",
        "\n",
        "def gen_batch_norm():\n",
        "    x = tf.keras.layers.BatchNormalization(epsilon=1e-5)\n",
        "    return x\n",
        "\n",
        "def get_generator_model():\n",
        "    row_input = tf.keras.layers.Input(shape=512)\n",
        "    x1 = gen_batch_norm()(row_input)\n",
        "    x1 = tf.keras.layers.Dense(512, activation=tf.nn.selu)(x1)\n",
        "    x = tf.keras.layers.Add()([x1, row_input])\n",
        "    x2 = gen_batch_norm()(x)\n",
        "    x2 = tf.keras.layers.Dense(512, activation=tf.nn.sigmoid)(x2)\n",
        "    x += tf.keras.layers.Add()([x2, x])\n",
        "\n",
        "    g_model = tf.keras.models.Model(row_input, x, name=\"generator\")\n",
        "    return g_model\n",
        "\n"
      ],
      "metadata": {
        "id": "KOavwPS1z6c8"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='5'>Discriminator definition</font>\n",
        "\n",
        "Input is with dimension 512 and has five dense layers of decreasing size 256, 128, 64, 32, 1\n",
        "\n",
        "Internal layers have Selu as activation function while the output layer has the linear function. The linear is necessary for the WGAN.\n",
        "\n",
        "Removed batch normalization from discriminator since it is not suggested for WGAN"
      ],
      "metadata": {
        "id": "qsh3Fefw0M0H"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# class Discriminator(tf.keras.Model):\n",
        "#     def __init__(self):\n",
        "#         super(Discriminator, self).__init__()\n",
        "#         self.D_DIMS = [256, 128, 64, 32]\n",
        "#         self.dense_layers = [tf.keras.layers.Dense(dim, activation=tf.nn.selu) for dim in self.D_DIMS]\n",
        "#         self.output_layer = tf.keras.layers.Dense(1, activation=tf.keras.activations.linear)\n",
        "\n",
        "#     def call(self, x):\n",
        "#         for h in range(len(self.D_DIMS)):\n",
        "#             x = self.dense_layers[h](x)\n",
        "#         x = self.output_layer(x)\n",
        "#         return x\n",
        "\n",
        "def disc_dense_layers(x):\n",
        "    x = tf.keras.layers.Dense(x, activation=tf.nn.selu)\n",
        "    return x\n",
        "\n",
        "def disc_output_layer(x):\n",
        "    x = tf.keras.layers.Dense(x, activation=tf.keras.activations.linear)\n",
        "    return x\n",
        "\n",
        "discriminator = tf.keras.Sequential(\n",
        "    [\n",
        "        tf.keras.layers.InputLayer((18)),\n",
        "        disc_dense_layers(256),\n",
        "        disc_dense_layers(128),\n",
        "        disc_dense_layers(64),\n",
        "        disc_dense_layers(32),\n",
        "        disc_output_layer(1),\n",
        "    ],\n",
        "    name=\"discriminator\",\n",
        ")"
      ],
      "metadata": {
        "id": "CVU-gPqB0MRM"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Additional parameters\n",
        "\n",
        "Initialize the model for the encoder, decoder, generator and discriminator\n",
        "\n",
        "Create the checkpoint manager and restore the trained autoencoder."
      ],
      "metadata": {
        "id": "B-bkdtdEy6NY"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "generator = get_generator_model()\n",
        "\n",
        "checkpoint_directory = \"/content/AE-WGAN-GP/WGAN-GP_checkpoints\"\n",
        "\n",
        "checkpoint = tf.train.Checkpoint(generator=generator, discriminator=discriminator)\n",
        "manager = tf.train.CheckpointManager(checkpoint, directory=checkpoint_directory, max_to_keep=5)\n",
        "\n",
        "checkpoint_ae.restore(manager_ae.latest_checkpoint).expect_partial().assert_consumed()\n"
      ],
      "metadata": {
        "id": "MATkYYapyyBI",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "7ef881cc-6e54-48bf-ea23-2cb634824eae"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.checkpoint.checkpoint.CheckpointLoadStatus at 0x7e3174c9c820>"
            ]
          },
          "metadata": {},
          "execution_count": 96
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='4'>Optimizer, main loss and decoder</font>\n",
        "\n",
        "The decoder is not trainable now\n",
        "\n",
        "Define the optimizer as Adam with relative low learning rate and very low betas. Added weigth decay to improve regularization (similar to l2)"
      ],
      "metadata": {
        "id": "OQeqPkZB2b54"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "decoder.trainable = False\n",
        "\n",
        "generator_optimizer = tf.keras.optimizers.AdamW(learning_rate=0.00005, weight_decay=0.0002,\n",
        "                                                    beta_1=0, beta_2=0.9)\n",
        "discriminator_optimizer = tf.keras.optimizers.AdamW(learning_rate=0.00005, weight_decay=0.0002,\n",
        "                                                        beta_1=0, beta_2=0.9)"
      ],
      "metadata": {
        "id": "9YsNcLVx2cXE"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='4'>Define the gradient penalty (GP)</font>"
      ],
      "metadata": {
        "id": "SwU45sK1024_"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def gradient_penalty(batch_size, real_data, fake_data):\n",
        "    \"\"\"Calculates the gradient penalty.\n",
        "\n",
        "    This loss is calculated on an interpolated data row\n",
        "    and added to the discriminator loss.\n",
        "    \"\"\"\n",
        "    # Get the interpolated data row\n",
        "    real_data = tf.cast(real_data, tf.float32)\n",
        "    alpha = tf.random.uniform([batch_size, 1], 0.0, 1.0)\n",
        "    diff = real_data - fake_data\n",
        "    interpolated = fake_data + alpha * diff\n",
        "\n",
        "    with tf.GradientTape() as gp_tape:\n",
        "        gp_tape.watch(interpolated)\n",
        "        # 1. Get the discriminator output for this interpolated data row.\n",
        "        pred = discriminator(interpolated, training=True)\n",
        "\n",
        "    # 2. Calculate the gradients w.r.t to this interpolated data row.\n",
        "    grads = gp_tape.gradient(pred, [interpolated])[0]\n",
        "    # 3. Calculate the norm of the gradients.\n",
        "    norm = tf.sqrt(tf.reduce_sum(tf.square(grads), axis=1))\n",
        "    gp = tf.reduce_mean((norm - 1.0) ** 2)\n",
        "    return gp"
      ],
      "metadata": {
        "id": "9tbmAjdE03NM"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Training steps\n",
        "\n",
        "### Define the training step for the discriminator/critic"
      ],
      "metadata": {
        "id": "dVLKXgMJ2Ae5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "@tf.function\n",
        "def d_step(real):\n",
        "    z = tf.random.normal(shape=[batchsize, Z_DIM])\n",
        "    gp_weight = 10\n",
        "\n",
        "    with tf.GradientTape() as disc_tape:\n",
        "        synthetic = decoder(generator(z, training=False))\n",
        "\n",
        "        real_output = discriminator(real)\n",
        "        fake_output = discriminator(synthetic)\n",
        "\n",
        "        disc_cost = tf.reduce_mean(fake_output) - tf.reduce_mean(real_output)\n",
        "\n",
        "        gp = gradient_penalty(batchsize, real, synthetic)\n",
        "\n",
        "        disc_loss = disc_cost + gp * gp_weight\n",
        "\n",
        "    gradients_of_discriminator = disc_tape.gradient(disc_loss, discriminator.trainable_variables)\n",
        "    discriminator_optimizer.apply_gradients(zip(gradients_of_discriminator, discriminator.trainable_variables))\n",
        "    return disc_loss"
      ],
      "metadata": {
        "id": "O23pXbqi2AWH"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='4'>Define the training step for the generator</font>"
      ],
      "metadata": {
        "id": "Fb_5HtLK2RdY"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "@tf.function\n",
        "def g_step():\n",
        "    z = tf.random.normal(shape=[batchsize, Z_DIM])\n",
        "    with tf.GradientTape() as gen_tape:\n",
        "        synthetic = decoder(generator(z, training=True))\n",
        "\n",
        "        if np.random.uniform(size=1) > 0.993:\n",
        "            print(\"synthetic\")\n",
        "            print(tf.round(synthetic).numpy())\n",
        "\n",
        "        fake_output = discriminator(synthetic)\n",
        "\n",
        "        gen_loss = -tf.reduce_mean(fake_output)\n",
        "\n",
        "    gradients_of_generator = gen_tape.gradient(gen_loss,\n",
        "                                               generator.trainable_variables + decoder.non_trainable_variables)\n",
        "    generator_optimizer.apply_gradients(\n",
        "        zip(gradients_of_generator, generator.trainable_variables + decoder.non_trainable_variables))\n",
        "    return gen_loss"
      ],
      "metadata": {
        "id": "7iYYep8e2RSU"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Train the model\n",
        "\n",
        "Load the data again in case autoencoder was run in a previous session and to retain only train data (don't need test data anymore)"
      ],
      "metadata": {
        "id": "r9hVubUWx-vK"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "train_data = np.load(\"/content/AE-WGAN-GP/Input_data/D3_events_ALL_OUTCOMES_ML.npy\", allow_pickle=True)\n",
        "dataset_train = tf.data.Dataset.from_tensor_slices(train_data).shuffle(\n",
        "        1000, reshuffle_each_iteration=True).batch(batchsize, drop_remainder=True)"
      ],
      "metadata": {
        "id": "aRipy0K-foUO"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Train the generator and discriminator/critic\n",
        "\n",
        "Save the checkpoint"
      ],
      "metadata": {
        "id": "oEv_Nyej1O-A"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "if train_WGAN_GP:\n",
        "    dloss_list = list()\n",
        "    gloss_list = list()\n",
        "    steps = len(dataset_train)\n",
        "\n",
        "    if train_from_partial:\n",
        "        checkpoint.restore(manager.latest_checkpoint).assert_consumed()\n",
        "        with open(os.path.join(checkpoint_directory, \"dloss.pkl\"), 'rb') as f:\n",
        "            dloss_list = pickle.load(f)\n",
        "        with open(os.path.join(checkpoint_directory, \"gloss.pkl\"), 'rb') as f:\n",
        "            gloss_list = pickle.load(f)\n",
        "\n",
        "    for epoch in range(20000):\n",
        "        start_time = time.time()\n",
        "        dloss = 0.0\n",
        "        gloss = 0.0\n",
        "        d_iter = 5\n",
        "        for batch_sample in dataset_train:\n",
        "            for _ in range(d_iter):\n",
        "                dloss += d_step(batch_sample)\n",
        "            gloss += g_step()\n",
        "\n",
        "        dloss_list.append(round(dloss.numpy() / (steps * d_iter * batchsize) * 10000, 1))\n",
        "        gloss_list.append(round(gloss.numpy() / steps / batchsize * 10000, 1))\n",
        "\n",
        "        duration_epoch = time.time() - start_time\n",
        "        format_str = 'epoch: %d, dloss = %.1f, gloss = %.1f, total_loss = %.1f (%.2f)'\n",
        "        print(format_str % (epoch, dloss / (steps * d_iter * batchsize) * 10000, gloss / steps / batchsize * 10000,\n",
        "         (dloss + gloss) / steps / batchsize * 1000, duration_epoch))\n",
        "\n",
        "    manager.save()\n",
        "    with open(os.path.join(checkpoint_directory, \"dloss.pkl\"), 'wb') as f:\n",
        "        pickle.dump(dloss_list, f)\n",
        "    with open(os.path.join(checkpoint_directory, \"gloss.pkl\"), 'wb') as f:\n",
        "        pickle.dump(gloss_list, f)"
      ],
      "metadata": {
        "id": "Mv7KKLVt1PTU",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "b764dd96-1f27-4b7e-9c7f-5dec3dfbf734"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[1;30;43mStreaming output truncated to the last 5000 lines.\u001b[0m\n",
            "epoch: 15000, dloss = 0.0, gloss = -2383.6, total_loss = -238.3 (0.69)\n",
            "epoch: 15001, dloss = -0.5, gloss = -2298.1, total_loss = -230.1 (0.68)\n",
            "epoch: 15002, dloss = -3.1, gloss = -2284.6, total_loss = -230.0 (0.70)\n",
            "epoch: 15003, dloss = 1.4, gloss = -2385.5, total_loss = -237.9 (0.69)\n",
            "epoch: 15004, dloss = 1.7, gloss = -2347.6, total_loss = -233.9 (0.82)\n",
            "epoch: 15005, dloss = 1.1, gloss = -2396.4, total_loss = -239.1 (0.88)\n",
            "epoch: 15006, dloss = 3.2, gloss = -2302.6, total_loss = -228.7 (0.69)\n",
            "epoch: 15007, dloss = 0.8, gloss = -2261.0, total_loss = -225.7 (0.69)\n",
            "epoch: 15008, dloss = -6.4, gloss = -2310.4, total_loss = -234.3 (0.67)\n",
            "epoch: 15009, dloss = -4.8, gloss = -2389.9, total_loss = -241.4 (0.70)\n",
            "epoch: 15010, dloss = 1.3, gloss = -2347.1, total_loss = -234.0 (0.69)\n",
            "epoch: 15011, dloss = -2.0, gloss = -2374.0, total_loss = -238.4 (0.69)\n",
            "epoch: 15012, dloss = 1.2, gloss = -2349.4, total_loss = -234.4 (0.70)\n",
            "epoch: 15013, dloss = -0.7, gloss = -2281.6, total_loss = -228.5 (0.69)\n",
            "epoch: 15014, dloss = 1.9, gloss = -2331.8, total_loss = -232.2 (0.67)\n",
            "epoch: 15015, dloss = -2.9, gloss = -2297.7, total_loss = -231.2 (0.67)\n",
            "epoch: 15016, dloss = -4.1, gloss = -2343.8, total_loss = -236.4 (0.67)\n",
            "epoch: 15017, dloss = 0.8, gloss = -2283.7, total_loss = -228.0 (0.69)\n",
            "epoch: 15018, dloss = -1.3, gloss = -2297.5, total_loss = -230.4 (0.66)\n",
            "epoch: 15019, dloss = 3.5, gloss = -2233.0, total_loss = -221.5 (0.69)\n",
            "epoch: 15020, dloss = -2.9, gloss = -2198.9, total_loss = -221.4 (0.92)\n",
            "epoch: 15021, dloss = 2.1, gloss = -2396.5, total_loss = -238.6 (0.79)\n",
            "epoch: 15022, dloss = 2.5, gloss = -2286.5, total_loss = -227.4 (0.68)\n",
            "epoch: 15023, dloss = -0.5, gloss = -2279.8, total_loss = -228.2 (0.66)\n",
            "epoch: 15024, dloss = 3.4, gloss = -2327.4, total_loss = -231.0 (0.66)\n",
            "epoch: 15025, dloss = -3.2, gloss = -2299.2, total_loss = -231.5 (0.69)\n",
            "epoch: 15026, dloss = 3.2, gloss = -2237.5, total_loss = -222.1 (0.68)\n",
            "epoch: 15027, dloss = -1.3, gloss = -2272.2, total_loss = -227.9 (0.67)\n",
            "epoch: 15028, dloss = -2.1, gloss = -2340.5, total_loss = -235.1 (0.68)\n",
            "epoch: 15029, dloss = 0.5, gloss = -2325.1, total_loss = -232.3 (0.70)\n",
            "epoch: 15030, dloss = 1.7, gloss = -2212.6, total_loss = -220.4 (0.72)\n",
            "epoch: 15031, dloss = -0.4, gloss = -2247.0, total_loss = -224.9 (0.72)\n",
            "epoch: 15032, dloss = 1.2, gloss = -2269.5, total_loss = -226.3 (0.71)\n",
            "epoch: 15033, dloss = -2.6, gloss = -2308.7, total_loss = -232.2 (0.69)\n",
            "epoch: 15034, dloss = 3.3, gloss = -2315.7, total_loss = -229.9 (0.71)\n",
            "epoch: 15035, dloss = 1.3, gloss = -2286.0, total_loss = -228.0 (0.78)\n",
            "epoch: 15036, dloss = 1.9, gloss = -2246.6, total_loss = -223.7 (0.97)\n",
            "epoch: 15037, dloss = 0.8, gloss = -2186.2, total_loss = -218.2 (0.67)\n",
            "epoch: 15038, dloss = -2.2, gloss = -2298.0, total_loss = -230.9 (0.69)\n",
            "epoch: 15039, dloss = 1.3, gloss = -2406.5, total_loss = -240.0 (0.69)\n",
            "epoch: 15040, dloss = -0.1, gloss = -2376.1, total_loss = -237.7 (0.67)\n",
            "epoch: 15041, dloss = -3.2, gloss = -2330.7, total_loss = -234.6 (0.66)\n",
            "epoch: 15042, dloss = -4.5, gloss = -2356.3, total_loss = -237.9 (0.68)\n",
            "epoch: 15043, dloss = -4.0, gloss = -2317.8, total_loss = -233.8 (0.68)\n",
            "epoch: 15044, dloss = -0.8, gloss = -2297.2, total_loss = -230.1 (0.67)\n",
            "epoch: 15045, dloss = 1.2, gloss = -2367.4, total_loss = -236.1 (0.69)\n",
            "epoch: 15046, dloss = -0.8, gloss = -2357.8, total_loss = -236.2 (0.71)\n",
            "epoch: 15047, dloss = 5.2, gloss = -2414.7, total_loss = -238.9 (0.69)\n",
            "epoch: 15048, dloss = -1.5, gloss = -2325.0, total_loss = -233.2 (0.69)\n",
            "epoch: 15049, dloss = 1.3, gloss = -2233.5, total_loss = -222.7 (0.72)\n",
            "epoch: 15050, dloss = -2.1, gloss = -2207.0, total_loss = -221.8 (0.69)\n",
            "epoch: 15051, dloss = 0.9, gloss = -2245.9, total_loss = -224.2 (0.93)\n",
            "epoch: 15052, dloss = 3.8, gloss = -2243.9, total_loss = -222.5 (0.79)\n",
            "epoch: 15053, dloss = 0.9, gloss = -2194.3, total_loss = -219.0 (0.69)\n",
            "epoch: 15054, dloss = -4.7, gloss = -2207.6, total_loss = -223.1 (0.71)\n",
            "epoch: 15055, dloss = -2.6, gloss = -2196.1, total_loss = -220.9 (0.92)\n",
            "epoch: 15056, dloss = 2.6, gloss = -2329.3, total_loss = -231.7 (0.88)\n",
            "epoch: 15057, dloss = 2.1, gloss = -2321.0, total_loss = -231.1 (0.69)\n",
            "epoch: 15058, dloss = 3.4, gloss = -2300.7, total_loss = -228.3 (0.71)\n",
            "epoch: 15059, dloss = -0.5, gloss = -2373.8, total_loss = -237.6 (0.67)\n",
            "epoch: 15060, dloss = -7.4, gloss = -2268.5, total_loss = -230.5 (0.68)\n",
            "epoch: 15061, dloss = 11.8, gloss = -2387.1, total_loss = -232.8 (0.70)\n",
            "epoch: 15062, dloss = 0.8, gloss = -2318.3, total_loss = -231.4 (0.69)\n",
            "epoch: 15063, dloss = 2.7, gloss = -2359.8, total_loss = -234.6 (0.70)\n",
            "epoch: 15064, dloss = -3.3, gloss = -2322.2, total_loss = -233.9 (0.71)\n",
            "epoch: 15065, dloss = 3.8, gloss = -2210.0, total_loss = -219.1 (0.69)\n",
            "epoch: 15066, dloss = -4.1, gloss = -2359.9, total_loss = -238.1 (1.00)\n",
            "epoch: 15067, dloss = 1.9, gloss = -2363.0, total_loss = -235.4 (0.78)\n",
            "epoch: 15068, dloss = 1.5, gloss = -2326.7, total_loss = -231.9 (0.99)\n",
            "epoch: 15069, dloss = 0.0, gloss = -2339.7, total_loss = -234.0 (0.68)\n",
            "epoch: 15070, dloss = 1.8, gloss = -2328.9, total_loss = -232.0 (0.68)\n",
            "epoch: 15071, dloss = 0.5, gloss = -2308.6, total_loss = -230.6 (0.71)\n",
            "epoch: 15072, dloss = -0.4, gloss = -2299.0, total_loss = -230.1 (0.68)\n",
            "epoch: 15073, dloss = -7.7, gloss = -2337.7, total_loss = -237.6 (0.68)\n",
            "epoch: 15074, dloss = -3.6, gloss = -2345.5, total_loss = -236.4 (0.69)\n",
            "epoch: 15075, dloss = -8.9, gloss = -2353.4, total_loss = -239.8 (0.71)\n",
            "epoch: 15076, dloss = 0.5, gloss = -2366.8, total_loss = -236.4 (0.68)\n",
            "epoch: 15077, dloss = -2.6, gloss = -2259.6, total_loss = -227.3 (0.71)\n",
            "epoch: 15078, dloss = 1.9, gloss = -2317.9, total_loss = -230.8 (0.70)\n",
            "epoch: 15079, dloss = -0.4, gloss = -2343.8, total_loss = -234.6 (0.69)\n",
            "epoch: 15080, dloss = -1.3, gloss = -2322.9, total_loss = -232.9 (0.70)\n",
            "epoch: 15081, dloss = -1.1, gloss = -2378.7, total_loss = -238.4 (1.00)\n",
            "epoch: 15082, dloss = 3.5, gloss = -2228.7, total_loss = -221.1 (0.75)\n",
            "epoch: 15083, dloss = -0.5, gloss = -2299.8, total_loss = -230.2 (0.69)\n",
            "epoch: 15084, dloss = 8.6, gloss = -2371.5, total_loss = -232.8 (0.71)\n",
            "epoch: 15085, dloss = 3.8, gloss = -2354.0, total_loss = -233.5 (0.70)\n",
            "epoch: 15086, dloss = -5.2, gloss = -2385.3, total_loss = -241.1 (0.71)\n",
            "epoch: 15087, dloss = -3.9, gloss = -2353.7, total_loss = -237.3 (0.69)\n",
            "epoch: 15088, dloss = 9.4, gloss = -2381.5, total_loss = -233.4 (0.69)\n",
            "epoch: 15089, dloss = 0.0, gloss = -2441.4, total_loss = -244.1 (0.71)\n",
            "epoch: 15090, dloss = -2.3, gloss = -2365.7, total_loss = -237.7 (0.68)\n",
            "epoch: 15091, dloss = 2.2, gloss = -2334.7, total_loss = -232.4 (0.68)\n",
            "epoch: 15092, dloss = 0.5, gloss = -2206.1, total_loss = -220.4 (0.70)\n",
            "epoch: 15093, dloss = 0.5, gloss = -2277.6, total_loss = -227.5 (0.68)\n",
            "epoch: 15094, dloss = -1.2, gloss = -2271.9, total_loss = -227.8 (0.71)\n",
            "epoch: 15095, dloss = -0.5, gloss = -2385.7, total_loss = -238.8 (0.69)\n",
            "epoch: 15096, dloss = -0.1, gloss = -2328.4, total_loss = -232.9 (0.91)\n",
            "epoch: 15097, dloss = 0.8, gloss = -2339.4, total_loss = -233.5 (0.84)\n",
            "epoch: 15098, dloss = 2.3, gloss = -2345.3, total_loss = -233.4 (0.70)\n",
            "epoch: 15099, dloss = -7.2, gloss = -2317.6, total_loss = -235.4 (0.71)\n",
            "epoch: 15100, dloss = 2.0, gloss = -2422.3, total_loss = -241.3 (0.70)\n",
            "epoch: 15101, dloss = 3.3, gloss = -2325.3, total_loss = -230.9 (0.68)\n",
            "epoch: 15102, dloss = -4.4, gloss = -2314.7, total_loss = -233.7 (0.70)\n",
            "epoch: 15103, dloss = 1.6, gloss = -2304.6, total_loss = -229.6 (0.69)\n",
            "epoch: 15104, dloss = 0.8, gloss = -2353.4, total_loss = -234.9 (0.67)\n",
            "epoch: 15105, dloss = 1.2, gloss = -2347.2, total_loss = -234.1 (0.67)\n",
            "epoch: 15106, dloss = -0.8, gloss = -2372.4, total_loss = -237.6 (0.69)\n",
            "epoch: 15107, dloss = 1.4, gloss = -2309.2, total_loss = -230.2 (0.68)\n",
            "epoch: 15108, dloss = 2.4, gloss = -2383.6, total_loss = -237.2 (0.69)\n",
            "epoch: 15109, dloss = 0.2, gloss = -2243.1, total_loss = -224.2 (0.67)\n",
            "epoch: 15110, dloss = 1.2, gloss = -2316.8, total_loss = -231.1 (0.70)\n",
            "epoch: 15111, dloss = 1.2, gloss = -2268.5, total_loss = -226.3 (0.75)\n",
            "epoch: 15112, dloss = -0.5, gloss = -2300.0, total_loss = -230.3 (1.00)\n",
            "epoch: 15113, dloss = 0.4, gloss = -2337.2, total_loss = -233.5 (0.70)\n",
            "epoch: 15114, dloss = 3.1, gloss = -2334.1, total_loss = -231.9 (0.65)\n",
            "epoch: 15115, dloss = -2.3, gloss = -2335.3, total_loss = -234.7 (0.70)\n",
            "epoch: 15116, dloss = -0.9, gloss = -2367.8, total_loss = -237.2 (0.69)\n",
            "epoch: 15117, dloss = 0.5, gloss = -2316.0, total_loss = -231.4 (0.69)\n",
            "epoch: 15118, dloss = 15.7, gloss = -2328.9, total_loss = -225.0 (0.70)\n",
            "epoch: 15119, dloss = 2.0, gloss = -2324.3, total_loss = -231.4 (0.70)\n",
            "epoch: 15120, dloss = 0.0, gloss = -2263.2, total_loss = -226.3 (0.67)\n",
            "epoch: 15121, dloss = 1.3, gloss = -2223.0, total_loss = -221.6 (0.70)\n",
            "epoch: 15122, dloss = 4.4, gloss = -2341.5, total_loss = -232.0 (0.69)\n",
            "epoch: 15123, dloss = 3.3, gloss = -2389.3, total_loss = -237.3 (0.67)\n",
            "epoch: 15124, dloss = 0.8, gloss = -2316.2, total_loss = -231.2 (0.69)\n",
            "epoch: 15125, dloss = 1.9, gloss = -2295.9, total_loss = -228.7 (0.69)\n",
            "epoch: 15126, dloss = -2.4, gloss = -2356.2, total_loss = -236.8 (0.68)\n",
            "epoch: 15127, dloss = 6.6, gloss = -2346.5, total_loss = -231.4 (0.90)\n",
            "epoch: 15128, dloss = 2.7, gloss = -2371.3, total_loss = -235.8 (0.83)\n",
            "epoch: 15129, dloss = 1.1, gloss = -2402.9, total_loss = -239.7 (0.68)\n",
            "epoch: 15130, dloss = 2.4, gloss = -2455.1, total_loss = -244.3 (0.67)\n",
            "epoch: 15131, dloss = 3.6, gloss = -2433.9, total_loss = -241.6 (0.70)\n",
            "epoch: 15132, dloss = 0.8, gloss = -2408.1, total_loss = -240.4 (0.68)\n",
            "epoch: 15133, dloss = 2.8, gloss = -2325.6, total_loss = -231.1 (0.70)\n",
            "epoch: 15134, dloss = 11.9, gloss = -2347.5, total_loss = -228.8 (0.68)\n",
            "epoch: 15135, dloss = 3.2, gloss = -2424.8, total_loss = -240.9 (0.71)\n",
            "epoch: 15136, dloss = 16.0, gloss = -2355.5, total_loss = -227.6 (0.67)\n",
            "epoch: 15137, dloss = -2.0, gloss = -2311.2, total_loss = -232.1 (0.67)\n",
            "epoch: 15138, dloss = 0.4, gloss = -2389.4, total_loss = -238.8 (0.70)\n",
            "epoch: 15139, dloss = -0.4, gloss = -2392.5, total_loss = -239.4 (0.69)\n",
            "epoch: 15140, dloss = -3.5, gloss = -2367.0, total_loss = -238.5 (0.69)\n",
            "epoch: 15141, dloss = -0.4, gloss = -2349.6, total_loss = -235.2 (0.68)\n",
            "epoch: 15142, dloss = -0.1, gloss = -2344.4, total_loss = -234.5 (0.72)\n",
            "epoch: 15143, dloss = -5.5, gloss = -2359.6, total_loss = -238.7 (1.02)\n",
            "epoch: 15144, dloss = 20.4, gloss = -2388.2, total_loss = -228.6 (0.67)\n",
            "epoch: 15145, dloss = 5.7, gloss = -2426.8, total_loss = -239.8 (0.68)\n",
            "epoch: 15146, dloss = 7.1, gloss = -2355.1, total_loss = -231.9 (0.69)\n",
            "epoch: 15147, dloss = 1.4, gloss = -2437.5, total_loss = -243.1 (0.67)\n",
            "epoch: 15148, dloss = -1.8, gloss = -2403.2, total_loss = -241.2 (0.69)\n",
            "epoch: 15149, dloss = 1.9, gloss = -2435.7, total_loss = -242.6 (0.67)\n",
            "epoch: 15150, dloss = 9.7, gloss = -2495.9, total_loss = -244.7 (0.68)\n",
            "epoch: 15151, dloss = 3.9, gloss = -2419.6, total_loss = -240.0 (0.68)\n",
            "epoch: 15152, dloss = 0.4, gloss = -2434.6, total_loss = -243.3 (0.69)\n",
            "epoch: 15153, dloss = -3.0, gloss = -2360.4, total_loss = -237.5 (0.70)\n",
            "epoch: 15154, dloss = 2.0, gloss = -2319.5, total_loss = -231.0 (0.72)\n",
            "epoch: 15155, dloss = 0.2, gloss = -2381.4, total_loss = -238.0 (0.69)\n",
            "epoch: 15156, dloss = 1.1, gloss = -2350.7, total_loss = -234.5 (0.70)\n",
            "epoch: 15157, dloss = -0.9, gloss = -2354.0, total_loss = -235.9 (0.70)\n",
            "epoch: 15158, dloss = 0.6, gloss = -2300.8, total_loss = -229.8 (0.86)\n",
            "epoch: 15159, dloss = 3.6, gloss = -2353.7, total_loss = -233.6 (0.88)\n",
            "epoch: 15160, dloss = -0.9, gloss = -2341.1, total_loss = -234.6 (0.67)\n",
            "epoch: 15161, dloss = -4.2, gloss = -2379.9, total_loss = -240.1 (0.68)\n",
            "epoch: 15162, dloss = -5.5, gloss = -2321.1, total_loss = -234.9 (0.70)\n",
            "epoch: 15163, dloss = -2.5, gloss = -2399.2, total_loss = -241.2 (0.67)\n",
            "epoch: 15164, dloss = 12.8, gloss = -2369.7, total_loss = -230.6 (0.69)\n",
            "epoch: 15165, dloss = 0.3, gloss = -2271.5, total_loss = -227.0 (0.69)\n",
            "epoch: 15166, dloss = 7.7, gloss = -2306.4, total_loss = -226.8 (0.67)\n",
            "epoch: 15167, dloss = 1.3, gloss = -2265.5, total_loss = -225.9 (0.68)\n",
            "epoch: 15168, dloss = -1.7, gloss = -2420.1, total_loss = -242.9 (0.68)\n",
            "epoch: 15169, dloss = -1.5, gloss = -2280.3, total_loss = -228.8 (0.69)\n",
            "epoch: 15170, dloss = 1.6, gloss = -2379.7, total_loss = -237.2 (0.68)\n",
            "epoch: 15171, dloss = 2.2, gloss = -2416.2, total_loss = -240.5 (0.70)\n",
            "epoch: 15172, dloss = 0.2, gloss = -2389.9, total_loss = -238.9 (0.68)\n",
            "epoch: 15173, dloss = -2.6, gloss = -2377.7, total_loss = -239.1 (0.69)\n",
            "epoch: 15174, dloss = -0.4, gloss = -2333.1, total_loss = -233.5 (1.02)\n",
            "epoch: 15175, dloss = -5.5, gloss = -2324.7, total_loss = -235.2 (0.70)\n",
            "epoch: 15176, dloss = 1.0, gloss = -2447.9, total_loss = -244.3 (0.67)\n",
            "epoch: 15177, dloss = -1.1, gloss = -2279.8, total_loss = -228.5 (0.68)\n",
            "epoch: 15178, dloss = -2.8, gloss = -2316.3, total_loss = -233.0 (0.69)\n",
            "epoch: 15179, dloss = 1.7, gloss = -2308.5, total_loss = -230.0 (0.70)\n",
            "epoch: 15180, dloss = -1.5, gloss = -2406.7, total_loss = -241.4 (0.68)\n",
            "epoch: 15181, dloss = 1.1, gloss = -2318.4, total_loss = -231.3 (0.69)\n",
            "epoch: 15182, dloss = -3.6, gloss = -2322.8, total_loss = -234.1 (0.66)\n",
            "epoch: 15183, dloss = 0.2, gloss = -2264.1, total_loss = -226.3 (0.68)\n",
            "epoch: 15184, dloss = 2.5, gloss = -2204.2, total_loss = -219.2 (0.68)\n",
            "epoch: 15185, dloss = -1.5, gloss = -2163.6, total_loss = -217.1 (0.69)\n",
            "epoch: 15186, dloss = -3.0, gloss = -2288.6, total_loss = -230.4 (0.68)\n",
            "epoch: 15187, dloss = 0.5, gloss = -2333.8, total_loss = -233.1 (0.70)\n",
            "epoch: 15188, dloss = -0.1, gloss = -2295.5, total_loss = -229.6 (0.67)\n",
            "epoch: 15189, dloss = 3.0, gloss = -2315.2, total_loss = -230.0 (0.82)\n",
            "epoch: 15190, dloss = 0.7, gloss = -2434.1, total_loss = -243.1 (0.92)\n",
            "epoch: 15191, dloss = -0.3, gloss = -2317.3, total_loss = -231.9 (0.70)\n",
            "epoch: 15192, dloss = -0.6, gloss = -2351.0, total_loss = -235.4 (0.69)\n",
            "epoch: 15193, dloss = 0.4, gloss = -2279.5, total_loss = -227.8 (0.66)\n",
            "epoch: 15194, dloss = 0.4, gloss = -2285.2, total_loss = -228.3 (0.67)\n",
            "epoch: 15195, dloss = 1.5, gloss = -2309.6, total_loss = -230.2 (0.70)\n",
            "epoch: 15196, dloss = -0.5, gloss = -2273.8, total_loss = -227.6 (0.69)\n",
            "epoch: 15197, dloss = 0.6, gloss = -2381.3, total_loss = -237.8 (0.68)\n",
            "epoch: 15198, dloss = 0.2, gloss = -2349.0, total_loss = -234.8 (0.70)\n",
            "epoch: 15199, dloss = 0.3, gloss = -2261.2, total_loss = -226.0 (0.68)\n",
            "epoch: 15200, dloss = 0.8, gloss = -2228.4, total_loss = -222.4 (0.69)\n",
            "epoch: 15201, dloss = -1.8, gloss = -2298.1, total_loss = -230.7 (0.70)\n",
            "epoch: 15202, dloss = -0.7, gloss = -2328.1, total_loss = -233.2 (0.67)\n",
            "epoch: 15203, dloss = 0.0, gloss = -2310.5, total_loss = -231.0 (0.71)\n",
            "epoch: 15204, dloss = 1.4, gloss = -2334.9, total_loss = -232.8 (0.70)\n",
            "epoch: 15205, dloss = 1.3, gloss = -2320.9, total_loss = -231.4 (0.94)\n",
            "epoch: 15206, dloss = 0.8, gloss = -2266.6, total_loss = -226.3 (0.81)\n",
            "epoch: 15207, dloss = 0.1, gloss = -2265.7, total_loss = -226.5 (0.67)\n",
            "epoch: 15208, dloss = -0.7, gloss = -2311.3, total_loss = -231.5 (0.71)\n",
            "epoch: 15209, dloss = -1.0, gloss = -2353.9, total_loss = -235.9 (0.68)\n",
            "epoch: 15210, dloss = -0.7, gloss = -2363.4, total_loss = -236.7 (0.73)\n",
            "epoch: 15211, dloss = 0.9, gloss = -2318.2, total_loss = -231.3 (0.69)\n",
            "epoch: 15212, dloss = -6.3, gloss = -2233.9, total_loss = -226.5 (0.68)\n",
            "epoch: 15213, dloss = -1.7, gloss = -2359.5, total_loss = -236.8 (0.66)\n",
            "epoch: 15214, dloss = -3.1, gloss = -2445.7, total_loss = -246.1 (0.65)\n",
            "epoch: 15215, dloss = -0.1, gloss = -2334.0, total_loss = -233.4 (0.67)\n",
            "epoch: 15216, dloss = -4.1, gloss = -2303.6, total_loss = -232.4 (0.68)\n",
            "epoch: 15217, dloss = 1.2, gloss = -2307.7, total_loss = -230.2 (0.68)\n",
            "epoch: 15218, dloss = -1.9, gloss = -2300.4, total_loss = -231.0 (0.69)\n",
            "epoch: 15219, dloss = -2.8, gloss = -2306.1, total_loss = -232.0 (0.68)\n",
            "epoch: 15220, dloss = 3.2, gloss = -2297.8, total_loss = -228.2 (0.72)\n",
            "epoch: 15221, dloss = 66.2, gloss = -2234.5, total_loss = -190.4 (0.99)\n",
            "epoch: 15222, dloss = 2.4, gloss = -2254.9, total_loss = -224.3 (0.66)\n",
            "epoch: 15223, dloss = 3.0, gloss = -2243.9, total_loss = -222.9 (0.70)\n",
            "epoch: 15224, dloss = 2.4, gloss = -2255.3, total_loss = -224.3 (0.69)\n",
            "epoch: 15225, dloss = -2.0, gloss = -2242.7, total_loss = -225.3 (0.70)\n",
            "epoch: 15226, dloss = -2.9, gloss = -2282.3, total_loss = -229.7 (0.86)\n",
            "epoch: 15227, dloss = 1.0, gloss = -2208.9, total_loss = -220.4 (0.97)\n",
            "epoch: 15228, dloss = 0.0, gloss = -2224.3, total_loss = -222.4 (0.69)\n",
            "epoch: 15229, dloss = 1.2, gloss = -2306.7, total_loss = -230.1 (0.68)\n",
            "epoch: 15230, dloss = -2.8, gloss = -2215.4, total_loss = -223.0 (0.66)\n",
            "epoch: 15231, dloss = 4.1, gloss = -2175.0, total_loss = -215.4 (0.67)\n",
            "epoch: 15232, dloss = -0.0, gloss = -2234.2, total_loss = -223.4 (0.69)\n",
            "epoch: 15233, dloss = -2.3, gloss = -2222.6, total_loss = -223.4 (0.66)\n",
            "epoch: 15234, dloss = 2.3, gloss = -2269.5, total_loss = -225.8 (0.67)\n",
            "epoch: 15235, dloss = -3.9, gloss = -2173.8, total_loss = -219.4 (0.71)\n",
            "epoch: 15236, dloss = 2.0, gloss = -2232.4, total_loss = -222.3 (1.03)\n",
            "epoch: 15237, dloss = 0.2, gloss = -2252.1, total_loss = -225.1 (0.68)\n",
            "epoch: 15238, dloss = 1.3, gloss = -2359.2, total_loss = -235.3 (0.70)\n",
            "epoch: 15239, dloss = -3.1, gloss = -2318.2, total_loss = -233.3 (0.70)\n",
            "epoch: 15240, dloss = -0.6, gloss = -2350.1, total_loss = -235.3 (0.69)\n",
            "epoch: 15241, dloss = 1.1, gloss = -2333.7, total_loss = -232.8 (0.71)\n",
            "epoch: 15242, dloss = -7.1, gloss = -2342.8, total_loss = -237.9 (0.70)\n",
            "epoch: 15243, dloss = 3.3, gloss = -2268.8, total_loss = -225.2 (0.69)\n",
            "epoch: 15244, dloss = -6.7, gloss = -2191.7, total_loss = -222.5 (0.68)\n",
            "epoch: 15245, dloss = -3.4, gloss = -2349.2, total_loss = -236.6 (0.70)\n",
            "epoch: 15246, dloss = -2.1, gloss = -2340.0, total_loss = -235.0 (0.69)\n",
            "epoch: 15247, dloss = 1.4, gloss = -2261.9, total_loss = -225.5 (0.68)\n",
            "epoch: 15248, dloss = -1.5, gloss = -2359.3, total_loss = -236.7 (0.66)\n",
            "epoch: 15249, dloss = 0.6, gloss = -2287.3, total_loss = -228.4 (0.66)\n",
            "epoch: 15250, dloss = -3.0, gloss = -2288.8, total_loss = -230.4 (0.69)\n",
            "epoch: 15251, dloss = -4.1, gloss = -2260.6, total_loss = -228.1 (0.88)\n",
            "epoch: 15252, dloss = 1.3, gloss = -2243.2, total_loss = -223.7 (0.86)\n",
            "epoch: 15253, dloss = 1.2, gloss = -2335.0, total_loss = -232.9 (0.68)\n",
            "epoch: 15254, dloss = 2.7, gloss = -2231.4, total_loss = -221.8 (0.68)\n",
            "epoch: 15255, dloss = -4.9, gloss = -2263.3, total_loss = -228.8 (0.68)\n",
            "epoch: 15256, dloss = -1.8, gloss = -2405.7, total_loss = -241.5 (0.69)\n",
            "epoch: 15257, dloss = -5.7, gloss = -2344.1, total_loss = -237.3 (0.68)\n",
            "epoch: 15258, dloss = -1.3, gloss = -2304.3, total_loss = -231.1 (0.68)\n",
            "epoch: 15259, dloss = -3.7, gloss = -2375.0, total_loss = -239.4 (0.68)\n",
            "epoch: 15260, dloss = -0.1, gloss = -2236.4, total_loss = -223.7 (0.69)\n",
            "epoch: 15261, dloss = -2.0, gloss = -2257.8, total_loss = -226.8 (0.69)\n",
            "epoch: 15262, dloss = -0.8, gloss = -2309.7, total_loss = -231.4 (0.72)\n",
            "epoch: 15263, dloss = 0.1, gloss = -2323.8, total_loss = -232.3 (0.70)\n",
            "epoch: 15264, dloss = 1.4, gloss = -2156.2, total_loss = -214.9 (0.70)\n",
            "epoch: 15265, dloss = 1.6, gloss = -2113.0, total_loss = -210.5 (0.66)\n",
            "epoch: 15266, dloss = -2.6, gloss = -2213.3, total_loss = -222.7 (0.70)\n",
            "epoch: 15267, dloss = -1.6, gloss = -2244.1, total_loss = -225.2 (1.04)\n",
            "epoch: 15268, dloss = 1.2, gloss = -2139.6, total_loss = -213.3 (0.71)\n",
            "epoch: 15269, dloss = 1.6, gloss = -2137.3, total_loss = -212.9 (0.69)\n",
            "epoch: 15270, dloss = -0.7, gloss = -2255.4, total_loss = -225.9 (0.68)\n",
            "epoch: 15271, dloss = 0.9, gloss = -2286.0, total_loss = -228.2 (0.69)\n",
            "epoch: 15272, dloss = 0.0, gloss = -2155.9, total_loss = -215.6 (0.69)\n",
            "epoch: 15273, dloss = -1.1, gloss = -2161.3, total_loss = -216.7 (0.69)\n",
            "epoch: 15274, dloss = -1.5, gloss = -2265.8, total_loss = -227.3 (0.69)\n",
            "epoch: 15275, dloss = 0.8, gloss = -2242.9, total_loss = -223.9 (0.67)\n",
            "epoch: 15276, dloss = 80.5, gloss = -2320.8, total_loss = -191.8 (0.67)\n",
            "epoch: 15277, dloss = -0.5, gloss = -2261.5, total_loss = -226.4 (0.69)\n",
            "epoch: 15278, dloss = 0.1, gloss = -2314.2, total_loss = -231.4 (0.69)\n",
            "epoch: 15279, dloss = -2.7, gloss = -2260.1, total_loss = -227.3 (0.68)\n",
            "epoch: 15280, dloss = 2.1, gloss = -2244.2, total_loss = -223.4 (0.69)\n",
            "epoch: 15281, dloss = 0.4, gloss = -2260.4, total_loss = -225.8 (0.67)\n",
            "epoch: 15282, dloss = 4.8, gloss = -2249.4, total_loss = -222.5 (0.85)\n",
            "epoch: 15283, dloss = -2.1, gloss = -2203.5, total_loss = -221.4 (0.90)\n",
            "epoch: 15284, dloss = -3.4, gloss = -2199.6, total_loss = -221.7 (0.66)\n",
            "epoch: 15285, dloss = -1.0, gloss = -2198.5, total_loss = -220.3 (0.70)\n",
            "epoch: 15286, dloss = -2.6, gloss = -2097.3, total_loss = -211.0 (0.68)\n",
            "epoch: 15287, dloss = -4.7, gloss = -2194.1, total_loss = -221.7 (0.69)\n",
            "epoch: 15288, dloss = -0.0, gloss = -2238.2, total_loss = -223.8 (0.68)\n",
            "epoch: 15289, dloss = -1.2, gloss = -2181.7, total_loss = -218.8 (0.67)\n",
            "epoch: 15290, dloss = -2.3, gloss = -2319.8, total_loss = -233.1 (0.70)\n",
            "epoch: 15291, dloss = -1.8, gloss = -2166.6, total_loss = -217.6 (0.69)\n",
            "epoch: 15292, dloss = -5.9, gloss = -2112.9, total_loss = -214.2 (0.67)\n",
            "epoch: 15293, dloss = -1.9, gloss = -2304.9, total_loss = -231.4 (0.70)\n",
            "epoch: 15294, dloss = 2.6, gloss = -2327.7, total_loss = -231.5 (0.69)\n",
            "epoch: 15295, dloss = 3.3, gloss = -2230.3, total_loss = -221.4 (0.66)\n",
            "epoch: 15296, dloss = 1.9, gloss = -2368.5, total_loss = -235.9 (0.69)\n",
            "epoch: 15297, dloss = -0.5, gloss = -2357.8, total_loss = -236.1 (0.71)\n",
            "epoch: 15298, dloss = 2.8, gloss = -2180.4, total_loss = -216.6 (1.06)\n",
            "epoch: 15299, dloss = 1.0, gloss = -2161.2, total_loss = -215.6 (1.17)\n",
            "epoch: 15300, dloss = -1.0, gloss = -2173.3, total_loss = -217.8 (0.70)\n",
            "epoch: 15301, dloss = 1.9, gloss = -2234.1, total_loss = -222.5 (0.67)\n",
            "epoch: 15302, dloss = 0.6, gloss = -2235.1, total_loss = -223.2 (0.68)\n",
            "epoch: 15303, dloss = -1.6, gloss = -2327.4, total_loss = -233.6 (0.69)\n",
            "epoch: 15304, dloss = 2.0, gloss = -2383.5, total_loss = -237.4 (0.65)\n",
            "epoch: 15305, dloss = 0.4, gloss = -2290.0, total_loss = -228.8 (0.70)\n",
            "epoch: 15306, dloss = 2.6, gloss = -2298.0, total_loss = -228.5 (0.72)\n",
            "epoch: 15307, dloss = 0.0, gloss = -2284.9, total_loss = -228.5 (0.67)\n",
            "epoch: 15308, dloss = 2.0, gloss = -2194.2, total_loss = -218.4 (0.69)\n",
            "epoch: 15309, dloss = -2.2, gloss = -2231.5, total_loss = -224.2 (0.69)\n",
            "epoch: 15310, dloss = 0.7, gloss = -2341.6, total_loss = -233.8 (0.69)\n",
            "epoch: 15311, dloss = -4.7, gloss = -2235.1, total_loss = -225.8 (0.70)\n",
            "epoch: 15312, dloss = -3.6, gloss = -2350.3, total_loss = -236.8 (1.01)\n",
            "epoch: 15313, dloss = 1.2, gloss = -2326.5, total_loss = -232.1 (1.03)\n",
            "epoch: 15314, dloss = 2.4, gloss = -2339.5, total_loss = -232.8 (0.79)\n",
            "epoch: 15315, dloss = -0.6, gloss = -2270.0, total_loss = -227.3 (0.68)\n",
            "epoch: 15316, dloss = -2.3, gloss = -2301.9, total_loss = -231.4 (0.68)\n",
            "epoch: 15317, dloss = 0.7, gloss = -2269.2, total_loss = -226.5 (0.70)\n",
            "epoch: 15318, dloss = -2.9, gloss = -2191.8, total_loss = -220.6 (0.69)\n",
            "epoch: 15319, dloss = -0.8, gloss = -2370.3, total_loss = -237.4 (0.68)\n",
            "epoch: 15320, dloss = -2.2, gloss = -2283.7, total_loss = -229.5 (0.68)\n",
            "epoch: 15321, dloss = 0.0, gloss = -2242.5, total_loss = -224.2 (0.68)\n",
            "epoch: 15322, dloss = 3.9, gloss = -2388.8, total_loss = -236.9 (0.69)\n",
            "epoch: 15323, dloss = 0.1, gloss = -2318.6, total_loss = -231.8 (0.69)\n",
            "epoch: 15324, dloss = -6.1, gloss = -2237.4, total_loss = -226.8 (0.68)\n",
            "epoch: 15325, dloss = 4.7, gloss = -2373.7, total_loss = -235.0 (0.71)\n",
            "epoch: 15326, dloss = -2.7, gloss = -2306.6, total_loss = -232.0 (0.70)\n",
            "epoch: 15327, dloss = 1.4, gloss = -2218.4, total_loss = -221.1 (0.66)\n",
            "epoch: 15328, dloss = 1.5, gloss = -2316.9, total_loss = -231.0 (0.81)\n",
            "epoch: 15329, dloss = -2.9, gloss = -2272.0, total_loss = -228.6 (0.94)\n",
            "epoch: 15330, dloss = 0.4, gloss = -2269.8, total_loss = -226.8 (0.74)\n",
            "epoch: 15331, dloss = -1.4, gloss = -2250.0, total_loss = -225.7 (0.71)\n",
            "epoch: 15332, dloss = 0.6, gloss = -2258.7, total_loss = -225.6 (0.69)\n",
            "epoch: 15333, dloss = -1.6, gloss = -2257.7, total_loss = -226.6 (0.69)\n",
            "epoch: 15334, dloss = 1.9, gloss = -2253.5, total_loss = -224.4 (0.68)\n",
            "epoch: 15335, dloss = 0.8, gloss = -2300.0, total_loss = -229.6 (0.69)\n",
            "epoch: 15336, dloss = 2.6, gloss = -2128.8, total_loss = -211.6 (0.69)\n",
            "epoch: 15337, dloss = -2.9, gloss = -2220.0, total_loss = -223.5 (0.68)\n",
            "epoch: 15338, dloss = 0.2, gloss = -2247.3, total_loss = -224.6 (0.68)\n",
            "epoch: 15339, dloss = -4.9, gloss = -2202.2, total_loss = -222.7 (0.69)\n",
            "epoch: 15340, dloss = 0.7, gloss = -2273.5, total_loss = -227.0 (0.66)\n",
            "epoch: 15341, dloss = -0.3, gloss = -2336.9, total_loss = -233.8 (0.68)\n",
            "epoch: 15342, dloss = -0.0, gloss = -2314.1, total_loss = -231.4 (0.67)\n",
            "epoch: 15343, dloss = 6.5, gloss = -2214.3, total_loss = -218.2 (0.69)\n",
            "epoch: 15344, dloss = -0.4, gloss = -2244.5, total_loss = -224.7 (0.95)\n",
            "epoch: 15345, dloss = -1.5, gloss = -2251.9, total_loss = -225.9 (0.78)\n",
            "epoch: 15346, dloss = 2.2, gloss = -2274.3, total_loss = -226.3 (0.70)\n",
            "epoch: 15347, dloss = -0.5, gloss = -2224.4, total_loss = -222.7 (0.69)\n",
            "epoch: 15348, dloss = 4.6, gloss = -2240.4, total_loss = -221.7 (0.67)\n",
            "epoch: 15349, dloss = -1.7, gloss = -2297.5, total_loss = -230.6 (0.70)\n",
            "epoch: 15350, dloss = 6.8, gloss = -2293.1, total_loss = -225.9 (0.69)\n",
            "epoch: 15351, dloss = -3.6, gloss = -2287.6, total_loss = -230.6 (0.69)\n",
            "epoch: 15352, dloss = -1.4, gloss = -2356.8, total_loss = -236.4 (0.68)\n",
            "epoch: 15353, dloss = -3.4, gloss = -2320.3, total_loss = -233.7 (0.69)\n",
            "epoch: 15354, dloss = -2.5, gloss = -2371.5, total_loss = -238.4 (0.67)\n",
            "epoch: 15355, dloss = -2.8, gloss = -2250.1, total_loss = -226.4 (0.67)\n",
            "epoch: 15356, dloss = -0.1, gloss = -2331.4, total_loss = -233.2 (0.69)\n",
            "epoch: 15357, dloss = 0.5, gloss = -2326.5, total_loss = -232.4 (0.68)\n",
            "epoch: 15358, dloss = -4.2, gloss = -2256.6, total_loss = -227.8 (0.70)\n",
            "epoch: 15359, dloss = 1.2, gloss = -2251.3, total_loss = -224.5 (0.78)\n",
            "epoch: 15360, dloss = 0.8, gloss = -2238.2, total_loss = -223.4 (0.94)\n",
            "epoch: 15361, dloss = 0.5, gloss = -2212.5, total_loss = -221.0 (0.67)\n",
            "epoch: 15362, dloss = 2.4, gloss = -2205.1, total_loss = -219.3 (0.68)\n",
            "epoch: 15363, dloss = -1.2, gloss = -2243.5, total_loss = -225.0 (1.26)\n",
            "epoch: 15364, dloss = 0.6, gloss = -2331.7, total_loss = -232.9 (0.71)\n",
            "epoch: 15365, dloss = 0.9, gloss = -2315.6, total_loss = -231.1 (0.68)\n",
            "epoch: 15366, dloss = -1.0, gloss = -2257.1, total_loss = -226.2 (0.68)\n",
            "epoch: 15367, dloss = 0.1, gloss = -2359.5, total_loss = -235.9 (0.71)\n",
            "epoch: 15368, dloss = 2.4, gloss = -2331.4, total_loss = -231.9 (0.69)\n",
            "epoch: 15369, dloss = 4.3, gloss = -2220.3, total_loss = -219.9 (0.68)\n",
            "epoch: 15370, dloss = 1.2, gloss = -2265.9, total_loss = -226.0 (0.66)\n",
            "epoch: 15371, dloss = 2.1, gloss = -2346.5, total_loss = -233.6 (0.71)\n",
            "epoch: 15372, dloss = -0.0, gloss = -2322.4, total_loss = -232.3 (0.66)\n",
            "epoch: 15373, dloss = -0.0, gloss = -2360.2, total_loss = -236.0 (0.68)\n",
            "epoch: 15374, dloss = 5.0, gloss = -2303.5, total_loss = -227.8 (0.81)\n",
            "epoch: 15375, dloss = 3.1, gloss = -2309.6, total_loss = -229.4 (0.89)\n",
            "epoch: 15376, dloss = 1.6, gloss = -2331.9, total_loss = -232.4 (0.67)\n",
            "epoch: 15377, dloss = 5.0, gloss = -2334.8, total_loss = -231.0 (0.70)\n",
            "epoch: 15378, dloss = -1.0, gloss = -2403.0, total_loss = -240.8 (0.70)\n",
            "epoch: 15379, dloss = -0.5, gloss = -2356.8, total_loss = -235.9 (0.69)\n",
            "epoch: 15380, dloss = 1.5, gloss = -2337.6, total_loss = -233.0 (0.70)\n",
            "epoch: 15381, dloss = -1.1, gloss = -2278.3, total_loss = -228.4 (0.67)\n",
            "epoch: 15382, dloss = -1.0, gloss = -2282.7, total_loss = -228.8 (0.68)\n",
            "epoch: 15383, dloss = 1.1, gloss = -2306.7, total_loss = -230.1 (0.70)\n",
            "epoch: 15384, dloss = 0.9, gloss = -2248.6, total_loss = -224.4 (0.70)\n",
            "epoch: 15385, dloss = -1.5, gloss = -2162.7, total_loss = -217.0 (0.68)\n",
            "epoch: 15386, dloss = 0.6, gloss = -2254.8, total_loss = -225.2 (0.69)\n",
            "epoch: 15387, dloss = 2.3, gloss = -2307.9, total_loss = -229.6 (0.68)\n",
            "epoch: 15388, dloss = 0.4, gloss = -2259.3, total_loss = -225.7 (0.69)\n",
            "epoch: 15389, dloss = -0.2, gloss = -2241.9, total_loss = -224.3 (0.69)\n",
            "epoch: 15390, dloss = -12.1, gloss = -2176.9, total_loss = -223.7 (0.98)\n",
            "epoch: 15391, dloss = -0.6, gloss = -2336.5, total_loss = -234.0 (0.77)\n",
            "epoch: 15392, dloss = -1.7, gloss = -2373.3, total_loss = -238.2 (0.70)\n",
            "epoch: 15393, dloss = -0.1, gloss = -2315.3, total_loss = -231.6 (0.68)\n",
            "epoch: 15394, dloss = 10.3, gloss = -2356.3, total_loss = -230.5 (0.69)\n",
            "epoch: 15395, dloss = -3.2, gloss = -2357.0, total_loss = -237.3 (0.70)\n",
            "epoch: 15396, dloss = 3.4, gloss = -2365.0, total_loss = -234.8 (0.70)\n",
            "epoch: 15397, dloss = -1.0, gloss = -2276.3, total_loss = -228.1 (0.69)\n",
            "epoch: 15398, dloss = 2.0, gloss = -2242.5, total_loss = -223.3 (0.68)\n",
            "epoch: 15399, dloss = 2.6, gloss = -2203.7, total_loss = -219.1 (0.68)\n",
            "epoch: 15400, dloss = 1.1, gloss = -2230.5, total_loss = -222.5 (0.67)\n",
            "epoch: 15401, dloss = -3.1, gloss = -2146.1, total_loss = -216.2 (0.69)\n",
            "epoch: 15402, dloss = -2.5, gloss = -2254.4, total_loss = -226.7 (0.68)\n",
            "epoch: 15403, dloss = 1.8, gloss = -2296.1, total_loss = -228.7 (0.69)\n",
            "epoch: 15404, dloss = 0.4, gloss = -2302.1, total_loss = -230.0 (0.68)\n",
            "epoch: 15405, dloss = 0.7, gloss = -2212.7, total_loss = -220.9 (0.84)\n",
            "epoch: 15406, dloss = 2.5, gloss = -2264.3, total_loss = -225.2 (0.95)\n",
            "epoch: 15407, dloss = 1.2, gloss = -2282.2, total_loss = -227.6 (0.68)\n",
            "epoch: 15408, dloss = 0.7, gloss = -2224.7, total_loss = -222.1 (0.66)\n",
            "epoch: 15409, dloss = -0.4, gloss = -2314.8, total_loss = -231.7 (0.69)\n",
            "epoch: 15410, dloss = 4.2, gloss = -2293.9, total_loss = -227.3 (0.69)\n",
            "epoch: 15411, dloss = -1.4, gloss = -2282.2, total_loss = -228.9 (0.70)\n",
            "epoch: 15412, dloss = -0.3, gloss = -2239.4, total_loss = -224.1 (0.67)\n",
            "epoch: 15413, dloss = 0.0, gloss = -2229.1, total_loss = -222.9 (0.72)\n",
            "epoch: 15414, dloss = 3.6, gloss = -2266.9, total_loss = -224.9 (0.70)\n",
            "epoch: 15415, dloss = 2.4, gloss = -2323.0, total_loss = -231.1 (0.66)\n",
            "epoch: 15416, dloss = -0.1, gloss = -2160.1, total_loss = -216.1 (0.69)\n",
            "epoch: 15417, dloss = 2.5, gloss = -2166.4, total_loss = -215.4 (0.71)\n",
            "epoch: 15418, dloss = 2.1, gloss = -2129.0, total_loss = -211.9 (0.98)\n",
            "epoch: 15419, dloss = 1.4, gloss = -2164.4, total_loss = -215.8 (1.16)\n",
            "epoch: 15420, dloss = 2.1, gloss = -2224.7, total_loss = -221.4 (1.33)\n",
            "epoch: 15421, dloss = -1.2, gloss = -2220.3, total_loss = -222.6 (0.69)\n",
            "epoch: 15422, dloss = 3.1, gloss = -2169.0, total_loss = -215.4 (0.69)\n",
            "epoch: 15423, dloss = -2.9, gloss = -2185.8, total_loss = -220.0 (0.68)\n",
            "epoch: 15424, dloss = -2.4, gloss = -2293.2, total_loss = -230.5 (0.68)\n",
            "epoch: 15425, dloss = 4.4, gloss = -2392.1, total_loss = -237.0 (0.68)\n",
            "epoch: 15426, dloss = 1.0, gloss = -2331.9, total_loss = -232.7 (0.67)\n",
            "epoch: 15427, dloss = 0.4, gloss = -2208.0, total_loss = -220.6 (0.67)\n",
            "epoch: 15428, dloss = -1.3, gloss = -2220.2, total_loss = -222.7 (0.66)\n",
            "epoch: 15429, dloss = 2.4, gloss = -2144.3, total_loss = -213.2 (0.68)\n",
            "epoch: 15430, dloss = -0.7, gloss = -2184.7, total_loss = -218.8 (0.68)\n",
            "epoch: 15431, dloss = -0.2, gloss = -2258.6, total_loss = -226.0 (0.70)\n",
            "epoch: 15432, dloss = -1.5, gloss = -2164.0, total_loss = -217.1 (0.68)\n",
            "epoch: 15433, dloss = 0.6, gloss = -2083.4, total_loss = -208.0 (0.67)\n",
            "epoch: 15434, dloss = -2.7, gloss = -2131.8, total_loss = -214.5 (0.70)\n",
            "epoch: 15435, dloss = 2.1, gloss = -2167.7, total_loss = -215.7 (0.83)\n",
            "epoch: 15436, dloss = -5.1, gloss = -2171.2, total_loss = -219.7 (0.88)\n",
            "epoch: 15437, dloss = -0.8, gloss = -2151.1, total_loss = -215.5 (0.68)\n",
            "epoch: 15438, dloss = -5.1, gloss = -2117.5, total_loss = -214.3 (0.69)\n",
            "epoch: 15439, dloss = -5.1, gloss = -2136.4, total_loss = -216.2 (0.66)\n",
            "epoch: 15440, dloss = -0.7, gloss = -2237.5, total_loss = -224.1 (0.69)\n",
            "epoch: 15441, dloss = -1.8, gloss = -2162.2, total_loss = -217.1 (0.67)\n",
            "epoch: 15442, dloss = -0.7, gloss = -2176.0, total_loss = -217.9 (0.67)\n",
            "epoch: 15443, dloss = 3.9, gloss = -2177.9, total_loss = -215.8 (0.68)\n",
            "epoch: 15444, dloss = 2.0, gloss = -2258.9, total_loss = -224.9 (0.70)\n",
            "epoch: 15445, dloss = 0.3, gloss = -2260.4, total_loss = -225.9 (0.67)\n",
            "epoch: 15446, dloss = 2.9, gloss = -2142.9, total_loss = -212.9 (0.68)\n",
            "epoch: 15447, dloss = -0.6, gloss = -2210.8, total_loss = -221.4 (0.70)\n",
            "epoch: 15448, dloss = 0.8, gloss = -2201.1, total_loss = -219.7 (0.69)\n",
            "epoch: 15449, dloss = 1.4, gloss = -2101.8, total_loss = -209.5 (0.69)\n",
            "epoch: 15450, dloss = -0.7, gloss = -2114.1, total_loss = -211.8 (0.70)\n",
            "epoch: 15451, dloss = 0.8, gloss = -2144.9, total_loss = -214.1 (0.99)\n",
            "epoch: 15452, dloss = -1.5, gloss = -2141.8, total_loss = -214.9 (0.76)\n",
            "epoch: 15453, dloss = -0.7, gloss = -2174.1, total_loss = -217.8 (0.67)\n",
            "epoch: 15454, dloss = -0.4, gloss = -2278.4, total_loss = -228.0 (0.68)\n",
            "epoch: 15455, dloss = -0.8, gloss = -2182.7, total_loss = -218.7 (0.68)\n",
            "epoch: 15456, dloss = 10.7, gloss = -2174.3, total_loss = -212.1 (0.68)\n",
            "epoch: 15457, dloss = 0.6, gloss = -2162.2, total_loss = -215.9 (0.68)\n",
            "epoch: 15458, dloss = 0.4, gloss = -2310.3, total_loss = -230.8 (0.68)\n",
            "epoch: 15459, dloss = -0.7, gloss = -2161.5, total_loss = -216.5 (0.68)\n",
            "epoch: 15460, dloss = -1.0, gloss = -2226.2, total_loss = -223.1 (0.69)\n",
            "epoch: 15461, dloss = -0.8, gloss = -2192.1, total_loss = -219.6 (0.69)\n",
            "epoch: 15462, dloss = -2.1, gloss = -2147.1, total_loss = -215.8 (0.69)\n",
            "epoch: 15463, dloss = 2.0, gloss = -2274.4, total_loss = -226.4 (0.69)\n",
            "epoch: 15464, dloss = 0.2, gloss = -2453.8, total_loss = -245.3 (0.68)\n",
            "epoch: 15465, dloss = -0.0, gloss = -2351.1, total_loss = -235.1 (0.68)\n",
            "epoch: 15466, dloss = 0.5, gloss = -2284.3, total_loss = -228.2 (0.78)\n",
            "epoch: 15467, dloss = 1.8, gloss = -2272.2, total_loss = -226.3 (0.94)\n",
            "epoch: 15468, dloss = -0.0, gloss = -2238.8, total_loss = -223.9 (0.69)\n",
            "epoch: 15469, dloss = 0.5, gloss = -2232.4, total_loss = -223.0 (0.71)\n",
            "epoch: 15470, dloss = 1.4, gloss = -2197.1, total_loss = -219.0 (0.70)\n",
            "epoch: 15471, dloss = 0.2, gloss = -2099.9, total_loss = -209.9 (0.67)\n",
            "epoch: 15472, dloss = 0.4, gloss = -2239.8, total_loss = -223.8 (0.69)\n",
            "epoch: 15473, dloss = -6.8, gloss = -2214.3, total_loss = -224.8 (0.68)\n",
            "epoch: 15474, dloss = 4.8, gloss = -2327.3, total_loss = -230.4 (0.68)\n",
            "epoch: 15475, dloss = -3.8, gloss = -2340.4, total_loss = -236.0 (0.70)\n",
            "epoch: 15476, dloss = -0.3, gloss = -2319.8, total_loss = -232.2 (0.71)\n",
            "epoch: 15477, dloss = 1.6, gloss = -2310.5, total_loss = -230.2 (0.68)\n",
            "epoch: 15478, dloss = 1.5, gloss = -2300.7, total_loss = -229.3 (0.68)\n",
            "epoch: 15479, dloss = 1.4, gloss = -2237.9, total_loss = -223.1 (0.69)\n",
            "epoch: 15480, dloss = 0.9, gloss = -2195.1, total_loss = -219.1 (0.69)\n",
            "epoch: 15481, dloss = 0.3, gloss = -2211.4, total_loss = -221.0 (0.69)\n",
            "epoch: 15482, dloss = 1.5, gloss = -2170.5, total_loss = -216.3 (0.97)\n",
            "epoch: 15483, dloss = 2.4, gloss = -2267.3, total_loss = -225.5 (0.76)\n",
            "epoch: 15484, dloss = -1.2, gloss = -2358.1, total_loss = -236.4 (0.69)\n",
            "epoch: 15485, dloss = 2.6, gloss = -2288.1, total_loss = -227.5 (0.69)\n",
            "epoch: 15486, dloss = -2.1, gloss = -2300.1, total_loss = -231.1 (0.69)\n",
            "epoch: 15487, dloss = 3.3, gloss = -2383.5, total_loss = -236.7 (0.68)\n",
            "epoch: 15488, dloss = -2.4, gloss = -2291.8, total_loss = -230.4 (0.67)\n",
            "epoch: 15489, dloss = -3.2, gloss = -2286.0, total_loss = -230.2 (0.70)\n",
            "epoch: 15490, dloss = 1.0, gloss = -2297.7, total_loss = -229.3 (0.71)\n",
            "epoch: 15491, dloss = 2.6, gloss = -2392.3, total_loss = -237.9 (0.69)\n",
            "epoch: 15492, dloss = -3.6, gloss = -2307.9, total_loss = -232.6 (0.70)\n",
            "epoch: 15493, dloss = -7.1, gloss = -2277.8, total_loss = -231.3 (0.69)\n",
            "epoch: 15494, dloss = 1.8, gloss = -2332.0, total_loss = -232.3 (0.69)\n",
            "epoch: 15495, dloss = -1.5, gloss = -2279.9, total_loss = -228.7 (0.70)\n",
            "epoch: 15496, dloss = -2.0, gloss = -2301.5, total_loss = -231.2 (0.69)\n",
            "epoch: 15497, dloss = 3.2, gloss = -2308.5, total_loss = -229.2 (0.81)\n",
            "epoch: 15498, dloss = -2.5, gloss = -2278.5, total_loss = -229.1 (0.92)\n",
            "epoch: 15499, dloss = 1.5, gloss = -2322.5, total_loss = -231.5 (0.68)\n",
            "epoch: 15500, dloss = 2.9, gloss = -2240.1, total_loss = -222.5 (0.68)\n",
            "epoch: 15501, dloss = 2.2, gloss = -2202.4, total_loss = -219.1 (0.68)\n",
            "epoch: 15502, dloss = -7.5, gloss = -2260.2, total_loss = -229.8 (0.68)\n",
            "epoch: 15503, dloss = -2.6, gloss = -2299.1, total_loss = -231.2 (0.68)\n",
            "epoch: 15504, dloss = 0.9, gloss = -2209.0, total_loss = -220.4 (0.68)\n",
            "epoch: 15505, dloss = 1.8, gloss = -2309.3, total_loss = -230.1 (0.68)\n",
            "epoch: 15506, dloss = -0.5, gloss = -2189.4, total_loss = -219.2 (0.67)\n",
            "epoch: 15507, dloss = -2.4, gloss = -2149.3, total_loss = -216.1 (0.69)\n",
            "epoch: 15508, dloss = -4.7, gloss = -2257.3, total_loss = -228.1 (0.68)\n",
            "epoch: 15509, dloss = 0.3, gloss = -2271.2, total_loss = -227.0 (0.85)\n",
            "epoch: 15510, dloss = 1.6, gloss = -2276.4, total_loss = -226.8 (0.97)\n",
            "epoch: 15511, dloss = 5.1, gloss = -2322.0, total_loss = -229.6 (0.70)\n",
            "epoch: 15512, dloss = -0.5, gloss = -2260.9, total_loss = -226.4 (0.83)\n",
            "epoch: 15513, dloss = -3.3, gloss = -2202.7, total_loss = -221.9 (0.92)\n",
            "epoch: 15514, dloss = 1.8, gloss = -2295.5, total_loss = -228.7 (0.70)\n",
            "epoch: 15515, dloss = -7.5, gloss = -2354.1, total_loss = -239.1 (0.70)\n",
            "epoch: 15516, dloss = -6.8, gloss = -2354.0, total_loss = -238.8 (0.69)\n",
            "epoch: 15517, dloss = 1.9, gloss = -2327.4, total_loss = -231.8 (0.71)\n",
            "epoch: 15518, dloss = 0.9, gloss = -2274.9, total_loss = -227.0 (0.69)\n",
            "epoch: 15519, dloss = -1.1, gloss = -2195.8, total_loss = -220.1 (0.70)\n",
            "epoch: 15520, dloss = -0.6, gloss = -2242.4, total_loss = -224.5 (0.70)\n",
            "epoch: 15521, dloss = -1.7, gloss = -2210.4, total_loss = -221.9 (0.69)\n",
            "epoch: 15522, dloss = 3.2, gloss = -2133.0, total_loss = -211.7 (0.69)\n",
            "epoch: 15523, dloss = -2.1, gloss = -2156.7, total_loss = -216.7 (0.68)\n",
            "epoch: 15524, dloss = -1.5, gloss = -2204.5, total_loss = -221.2 (0.69)\n",
            "epoch: 15525, dloss = -2.1, gloss = -2165.4, total_loss = -217.6 (0.67)\n",
            "epoch: 15526, dloss = -10.5, gloss = -2266.6, total_loss = -231.9 (0.69)\n",
            "epoch: 15527, dloss = -0.8, gloss = -2283.6, total_loss = -228.8 (0.71)\n",
            "epoch: 15528, dloss = -2.1, gloss = -2307.1, total_loss = -231.8 (1.27)\n",
            "epoch: 15529, dloss = 1.0, gloss = -2297.1, total_loss = -229.2 (0.70)\n",
            "epoch: 15530, dloss = 2.8, gloss = -2227.6, total_loss = -221.3 (0.69)\n",
            "epoch: 15531, dloss = 1.6, gloss = -2193.8, total_loss = -218.6 (0.71)\n",
            "epoch: 15532, dloss = 0.5, gloss = -2190.8, total_loss = -218.8 (0.70)\n",
            "epoch: 15533, dloss = -1.1, gloss = -2265.3, total_loss = -227.1 (0.69)\n",
            "epoch: 15534, dloss = -2.3, gloss = -2310.7, total_loss = -232.2 (0.69)\n",
            "epoch: 15535, dloss = -2.2, gloss = -2328.2, total_loss = -233.9 (0.69)\n",
            "epoch: 15536, dloss = 0.3, gloss = -2313.8, total_loss = -231.2 (0.69)\n",
            "epoch: 15537, dloss = 1.7, gloss = -2340.4, total_loss = -233.2 (0.69)\n",
            "epoch: 15538, dloss = 5.2, gloss = -2292.8, total_loss = -226.7 (0.67)\n",
            "epoch: 15539, dloss = -1.7, gloss = -2272.0, total_loss = -228.0 (0.68)\n",
            "epoch: 15540, dloss = -6.8, gloss = -2183.7, total_loss = -221.8 (0.68)\n",
            "epoch: 15541, dloss = -2.8, gloss = -2310.0, total_loss = -232.4 (0.67)\n",
            "epoch: 15542, dloss = 4.7, gloss = -2195.7, total_loss = -217.2 (0.67)\n",
            "epoch: 15543, dloss = 0.3, gloss = -2240.7, total_loss = -223.9 (0.93)\n",
            "epoch: 15544, dloss = -3.1, gloss = -2255.0, total_loss = -227.0 (0.97)\n",
            "epoch: 15545, dloss = -0.3, gloss = -2269.5, total_loss = -227.1 (0.89)\n",
            "epoch: 15546, dloss = -0.8, gloss = -2306.2, total_loss = -231.0 (0.69)\n",
            "epoch: 15547, dloss = 0.5, gloss = -2315.9, total_loss = -231.3 (0.69)\n",
            "epoch: 15548, dloss = 0.2, gloss = -2326.6, total_loss = -232.6 (0.69)\n",
            "epoch: 15549, dloss = -8.4, gloss = -2222.2, total_loss = -226.4 (0.68)\n",
            "epoch: 15550, dloss = 1.4, gloss = -2248.1, total_loss = -224.1 (0.68)\n",
            "epoch: 15551, dloss = -0.6, gloss = -2351.6, total_loss = -235.5 (0.70)\n",
            "epoch: 15552, dloss = -1.8, gloss = -2319.0, total_loss = -232.8 (0.68)\n",
            "epoch: 15553, dloss = 1.4, gloss = -2265.6, total_loss = -225.8 (0.68)\n",
            "epoch: 15554, dloss = 3.0, gloss = -2214.0, total_loss = -219.9 (0.68)\n",
            "epoch: 15555, dloss = 2.3, gloss = -2170.6, total_loss = -215.9 (0.68)\n",
            "epoch: 15556, dloss = 5.0, gloss = -2177.0, total_loss = -215.2 (0.83)\n",
            "epoch: 15557, dloss = -3.1, gloss = -2155.0, total_loss = -217.1 (0.99)\n",
            "epoch: 15558, dloss = 3.1, gloss = -2276.6, total_loss = -226.1 (0.96)\n",
            "epoch: 15559, dloss = -3.0, gloss = -2306.2, total_loss = -232.1 (0.67)\n",
            "epoch: 15560, dloss = -1.4, gloss = -2272.8, total_loss = -228.0 (0.69)\n",
            "epoch: 15561, dloss = 2.5, gloss = -2160.2, total_loss = -214.8 (0.69)\n",
            "epoch: 15562, dloss = 2.5, gloss = -2177.2, total_loss = -216.5 (0.68)\n",
            "epoch: 15563, dloss = -2.8, gloss = -2216.1, total_loss = -223.0 (0.68)\n",
            "epoch: 15564, dloss = -4.8, gloss = -2291.6, total_loss = -231.6 (0.69)\n",
            "epoch: 15565, dloss = 0.5, gloss = -2316.9, total_loss = -231.5 (0.70)\n",
            "epoch: 15566, dloss = -0.7, gloss = -2347.5, total_loss = -235.1 (0.71)\n",
            "epoch: 15567, dloss = 3.1, gloss = -2276.3, total_loss = -226.1 (0.70)\n",
            "epoch: 15568, dloss = -3.8, gloss = -2245.2, total_loss = -226.4 (0.69)\n",
            "epoch: 15569, dloss = 3.4, gloss = -2255.0, total_loss = -223.8 (0.71)\n",
            "epoch: 15570, dloss = 1.7, gloss = -2328.0, total_loss = -231.9 (0.72)\n",
            "epoch: 15571, dloss = -3.5, gloss = -2206.8, total_loss = -222.4 (0.68)\n",
            "epoch: 15572, dloss = -0.2, gloss = -2241.9, total_loss = -224.3 (0.68)\n",
            "epoch: 15573, dloss = -0.1, gloss = -2159.2, total_loss = -216.0 (0.98)\n",
            "epoch: 15574, dloss = 2.7, gloss = -2199.7, total_loss = -218.6 (0.77)\n",
            "epoch: 15575, dloss = -2.1, gloss = -2201.4, total_loss = -221.2 (0.67)\n",
            "epoch: 15576, dloss = 0.1, gloss = -2306.9, total_loss = -230.6 (0.68)\n",
            "epoch: 15577, dloss = 0.0, gloss = -2316.2, total_loss = -231.6 (0.71)\n",
            "epoch: 15578, dloss = 3.7, gloss = -2227.0, total_loss = -220.8 (0.69)\n",
            "epoch: 15579, dloss = 1.2, gloss = -2309.6, total_loss = -230.4 (0.66)\n",
            "epoch: 15580, dloss = 2.4, gloss = -2292.2, total_loss = -228.0 (0.70)\n",
            "epoch: 15581, dloss = 3.5, gloss = -2338.8, total_loss = -232.1 (0.69)\n",
            "epoch: 15582, dloss = 0.7, gloss = -2349.3, total_loss = -234.6 (0.68)\n",
            "epoch: 15583, dloss = -2.6, gloss = -2349.7, total_loss = -236.3 (0.69)\n",
            "epoch: 15584, dloss = -1.6, gloss = -2430.6, total_loss = -243.9 (0.69)\n",
            "epoch: 15585, dloss = 0.6, gloss = -2412.7, total_loss = -241.0 (0.69)\n",
            "epoch: 15586, dloss = 7.9, gloss = -2392.6, total_loss = -235.3 (0.69)\n",
            "epoch: 15587, dloss = -0.6, gloss = -2331.4, total_loss = -233.4 (0.69)\n",
            "epoch: 15588, dloss = -0.4, gloss = -2324.3, total_loss = -232.7 (0.76)\n",
            "epoch: 15589, dloss = -3.1, gloss = -2265.8, total_loss = -228.1 (0.96)\n",
            "epoch: 15590, dloss = 1.5, gloss = -2345.0, total_loss = -233.8 (0.70)\n",
            "epoch: 15591, dloss = 4.3, gloss = -2299.9, total_loss = -227.9 (0.69)\n",
            "epoch: 15592, dloss = -4.6, gloss = -2301.4, total_loss = -232.4 (0.69)\n",
            "epoch: 15593, dloss = 1.0, gloss = -2323.6, total_loss = -231.9 (0.69)\n",
            "epoch: 15594, dloss = 2.9, gloss = -2352.0, total_loss = -233.7 (0.70)\n",
            "epoch: 15595, dloss = -0.4, gloss = -2267.0, total_loss = -226.9 (0.68)\n",
            "epoch: 15596, dloss = -0.1, gloss = -2175.3, total_loss = -217.6 (0.68)\n",
            "epoch: 15597, dloss = 2.3, gloss = -2274.5, total_loss = -226.3 (0.69)\n",
            "epoch: 15598, dloss = -2.6, gloss = -2285.1, total_loss = -229.8 (0.70)\n",
            "epoch: 15599, dloss = 2.7, gloss = -2323.9, total_loss = -231.0 (0.67)\n",
            "epoch: 15600, dloss = 1.4, gloss = -2330.2, total_loss = -232.3 (0.72)\n",
            "epoch: 15601, dloss = -8.6, gloss = -2296.2, total_loss = -233.9 (0.68)\n",
            "epoch: 15602, dloss = 3.0, gloss = -2306.8, total_loss = -229.2 (0.68)\n",
            "epoch: 15603, dloss = 4.2, gloss = -2288.0, total_loss = -226.7 (0.69)\n",
            "epoch: 15604, dloss = 0.6, gloss = -2282.9, total_loss = -228.0 (0.94)\n",
            "epoch: 15605, dloss = -2.1, gloss = -2264.8, total_loss = -227.5 (0.78)\n",
            "epoch: 15606, dloss = -0.5, gloss = -2214.7, total_loss = -221.7 (0.70)\n",
            "epoch: 15607, dloss = -5.4, gloss = -2308.7, total_loss = -233.6 (0.67)\n",
            "epoch: 15608, dloss = 4.6, gloss = -2275.5, total_loss = -225.2 (0.67)\n",
            "epoch: 15609, dloss = -4.4, gloss = -2293.8, total_loss = -231.6 (0.71)\n",
            "epoch: 15610, dloss = -6.4, gloss = -2327.3, total_loss = -235.9 (0.71)\n",
            "epoch: 15611, dloss = -0.0, gloss = -2292.9, total_loss = -229.3 (0.70)\n",
            "epoch: 15612, dloss = 1.0, gloss = -2343.6, total_loss = -233.9 (0.70)\n",
            "epoch: 15613, dloss = 4.6, gloss = -2335.3, total_loss = -231.2 (0.67)\n",
            "epoch: 15614, dloss = -3.3, gloss = -2343.2, total_loss = -236.0 (0.71)\n",
            "epoch: 15615, dloss = -1.0, gloss = -2343.0, total_loss = -234.8 (0.68)\n",
            "epoch: 15616, dloss = 5.3, gloss = -2243.1, total_loss = -221.7 (0.70)\n",
            "epoch: 15617, dloss = 2.1, gloss = -2276.1, total_loss = -226.6 (0.68)\n",
            "epoch: 15618, dloss = 1.4, gloss = -2295.2, total_loss = -228.8 (0.72)\n",
            "epoch: 15619, dloss = 0.6, gloss = -2251.3, total_loss = -224.8 (0.75)\n",
            "epoch: 15620, dloss = 0.9, gloss = -2261.2, total_loss = -225.7 (0.96)\n",
            "epoch: 15621, dloss = 2.4, gloss = -2244.8, total_loss = -223.3 (0.70)\n",
            "epoch: 15622, dloss = 2.5, gloss = -2274.6, total_loss = -226.2 (0.69)\n",
            "epoch: 15623, dloss = 3.4, gloss = -2342.8, total_loss = -232.6 (0.68)\n",
            "epoch: 15624, dloss = 0.2, gloss = -2330.0, total_loss = -232.9 (0.70)\n",
            "epoch: 15625, dloss = -6.5, gloss = -2288.9, total_loss = -232.2 (0.66)\n",
            "epoch: 15626, dloss = -1.3, gloss = -2267.9, total_loss = -227.4 (0.72)\n",
            "epoch: 15627, dloss = 0.8, gloss = -2264.3, total_loss = -226.0 (0.68)\n",
            "epoch: 15628, dloss = 0.9, gloss = -2215.1, total_loss = -221.1 (0.67)\n",
            "epoch: 15629, dloss = -0.4, gloss = -2203.4, total_loss = -220.5 (0.68)\n",
            "epoch: 15630, dloss = 1.8, gloss = -2233.7, total_loss = -222.5 (0.69)\n",
            "epoch: 15631, dloss = -2.0, gloss = -2289.0, total_loss = -229.9 (0.67)\n",
            "epoch: 15632, dloss = 0.8, gloss = -2329.9, total_loss = -232.6 (0.68)\n",
            "epoch: 15633, dloss = 1.5, gloss = -2376.1, total_loss = -236.9 (0.69)\n",
            "epoch: 15634, dloss = 2.7, gloss = -2404.5, total_loss = -239.1 (0.70)\n",
            "epoch: 15635, dloss = 1.2, gloss = -2318.6, total_loss = -231.2 (0.88)\n",
            "epoch: 15636, dloss = 1.3, gloss = -2391.9, total_loss = -238.6 (0.82)\n",
            "epoch: 15637, dloss = 1.0, gloss = -2368.8, total_loss = -236.4 (0.67)\n",
            "epoch: 15638, dloss = 2.8, gloss = -2367.3, total_loss = -235.3 (0.68)\n",
            "epoch: 15639, dloss = 4.0, gloss = -2328.0, total_loss = -230.8 (0.67)\n",
            "epoch: 15640, dloss = 0.7, gloss = -2255.7, total_loss = -225.2 (0.68)\n",
            "epoch: 15641, dloss = -1.3, gloss = -2250.2, total_loss = -225.7 (0.68)\n",
            "epoch: 15642, dloss = 3.9, gloss = -2224.5, total_loss = -220.5 (0.69)\n",
            "epoch: 15643, dloss = 1.6, gloss = -2265.7, total_loss = -225.8 (0.68)\n",
            "epoch: 15644, dloss = -8.9, gloss = -2216.9, total_loss = -226.2 (0.67)\n",
            "epoch: 15645, dloss = 2.8, gloss = -2291.9, total_loss = -227.8 (0.69)\n",
            "epoch: 15646, dloss = -3.0, gloss = -2287.3, total_loss = -230.2 (0.68)\n",
            "epoch: 15647, dloss = 2.4, gloss = -2339.4, total_loss = -232.7 (0.68)\n",
            "epoch: 15648, dloss = 1.6, gloss = -2302.7, total_loss = -229.5 (0.70)\n",
            "epoch: 15649, dloss = 3.4, gloss = -2228.3, total_loss = -221.1 (0.69)\n",
            "epoch: 15650, dloss = -4.8, gloss = -2225.7, total_loss = -224.9 (0.69)\n",
            "epoch: 15651, dloss = 0.5, gloss = -2207.4, total_loss = -220.5 (1.26)\n",
            "epoch: 15652, dloss = 1.6, gloss = -2241.3, total_loss = -223.3 (0.69)\n",
            "epoch: 15653, dloss = 2.9, gloss = -2143.7, total_loss = -212.9 (0.72)\n",
            "epoch: 15654, dloss = 1.0, gloss = -2253.1, total_loss = -224.8 (0.71)\n",
            "epoch: 15655, dloss = 8.7, gloss = -2298.2, total_loss = -225.5 (0.68)\n",
            "epoch: 15656, dloss = -1.5, gloss = -2260.5, total_loss = -226.8 (0.69)\n",
            "epoch: 15657, dloss = 0.5, gloss = -2226.5, total_loss = -222.4 (0.69)\n",
            "epoch: 15658, dloss = 2.7, gloss = -2310.5, total_loss = -229.7 (0.70)\n",
            "epoch: 15659, dloss = 2.4, gloss = -2214.1, total_loss = -220.2 (0.69)\n",
            "epoch: 15660, dloss = -0.2, gloss = -2197.5, total_loss = -219.8 (0.70)\n",
            "epoch: 15661, dloss = 1.0, gloss = -2212.8, total_loss = -220.8 (0.69)\n",
            "epoch: 15662, dloss = 1.0, gloss = -2212.4, total_loss = -220.7 (0.69)\n",
            "epoch: 15663, dloss = 1.1, gloss = -2321.9, total_loss = -231.7 (0.68)\n",
            "epoch: 15664, dloss = -2.1, gloss = -2292.9, total_loss = -230.4 (0.65)\n",
            "epoch: 15665, dloss = 0.7, gloss = -2326.1, total_loss = -232.2 (0.67)\n",
            "epoch: 15666, dloss = -2.9, gloss = -2334.5, total_loss = -234.9 (0.97)\n",
            "epoch: 15667, dloss = -1.2, gloss = -2261.0, total_loss = -226.7 (0.76)\n",
            "epoch: 15668, dloss = 3.3, gloss = -2245.1, total_loss = -222.9 (0.65)\n",
            "epoch: 15669, dloss = -5.1, gloss = -2320.7, total_loss = -234.6 (0.69)\n",
            "epoch: 15670, dloss = 1.1, gloss = -2299.0, total_loss = -229.4 (0.67)\n",
            "epoch: 15671, dloss = -1.2, gloss = -2251.1, total_loss = -225.7 (0.68)\n",
            "epoch: 15672, dloss = 0.7, gloss = -2269.9, total_loss = -226.7 (0.68)\n",
            "epoch: 15673, dloss = 2.8, gloss = -2237.7, total_loss = -222.4 (0.69)\n",
            "epoch: 15674, dloss = -2.7, gloss = -2293.1, total_loss = -230.7 (0.70)\n",
            "epoch: 15675, dloss = 2.4, gloss = -2337.5, total_loss = -232.5 (0.69)\n",
            "epoch: 15676, dloss = -3.1, gloss = -2331.4, total_loss = -234.7 (0.70)\n",
            "epoch: 15677, dloss = 0.1, gloss = -2333.5, total_loss = -233.3 (0.68)\n",
            "epoch: 15678, dloss = 4.0, gloss = -2260.6, total_loss = -224.0 (0.68)\n",
            "epoch: 15679, dloss = -2.6, gloss = -2256.8, total_loss = -227.0 (0.68)\n",
            "epoch: 15680, dloss = 29.8, gloss = -2265.3, total_loss = -211.6 (0.68)\n",
            "epoch: 15681, dloss = 0.0, gloss = -2224.4, total_loss = -222.4 (0.72)\n",
            "epoch: 15682, dloss = 2.4, gloss = -2222.2, total_loss = -221.0 (1.03)\n",
            "epoch: 15683, dloss = 2.0, gloss = -2338.1, total_loss = -232.8 (0.70)\n",
            "epoch: 15684, dloss = 2.1, gloss = -2419.0, total_loss = -240.9 (0.70)\n",
            "epoch: 15685, dloss = 3.5, gloss = -2248.8, total_loss = -223.1 (0.69)\n",
            "epoch: 15686, dloss = -0.5, gloss = -2268.8, total_loss = -227.1 (0.68)\n",
            "epoch: 15687, dloss = -0.7, gloss = -2332.8, total_loss = -233.6 (0.70)\n",
            "epoch: 15688, dloss = 2.3, gloss = -2327.7, total_loss = -231.6 (0.70)\n",
            "epoch: 15689, dloss = 1.0, gloss = -2323.5, total_loss = -231.8 (0.68)\n",
            "epoch: 15690, dloss = 0.7, gloss = -2395.0, total_loss = -239.2 (0.71)\n",
            "epoch: 15691, dloss = 1.0, gloss = -2320.7, total_loss = -231.5 (0.69)\n",
            "epoch: 15692, dloss = -1.0, gloss = -2367.6, total_loss = -237.3 (0.68)\n",
            "epoch: 15693, dloss = -1.4, gloss = -2340.8, total_loss = -234.8 (0.69)\n",
            "epoch: 15694, dloss = 3.8, gloss = -2363.6, total_loss = -234.5 (0.67)\n",
            "epoch: 15695, dloss = 3.6, gloss = -2332.1, total_loss = -231.4 (0.71)\n",
            "epoch: 15696, dloss = 4.7, gloss = -2347.4, total_loss = -232.4 (0.69)\n",
            "epoch: 15697, dloss = 0.4, gloss = -2322.1, total_loss = -232.0 (0.93)\n",
            "epoch: 15698, dloss = -0.0, gloss = -2279.1, total_loss = -227.9 (0.81)\n",
            "epoch: 15699, dloss = 3.3, gloss = -2275.6, total_loss = -225.9 (0.71)\n",
            "epoch: 15700, dloss = -5.6, gloss = -2281.9, total_loss = -231.0 (0.69)\n",
            "epoch: 15701, dloss = -1.0, gloss = -2300.9, total_loss = -230.6 (0.67)\n",
            "epoch: 15702, dloss = -0.8, gloss = -2366.0, total_loss = -237.0 (0.70)\n",
            "epoch: 15703, dloss = 0.3, gloss = -2319.0, total_loss = -231.8 (0.67)\n",
            "epoch: 15704, dloss = 0.9, gloss = -2221.7, total_loss = -221.7 (0.70)\n",
            "epoch: 15705, dloss = 7.0, gloss = -2233.5, total_loss = -219.9 (0.69)\n",
            "epoch: 15706, dloss = -0.7, gloss = -2257.1, total_loss = -226.1 (0.68)\n",
            "epoch: 15707, dloss = 0.8, gloss = -2254.7, total_loss = -225.0 (0.69)\n",
            "epoch: 15708, dloss = -2.4, gloss = -2290.7, total_loss = -230.3 (0.68)\n",
            "epoch: 15709, dloss = 1.3, gloss = -2301.4, total_loss = -229.5 (0.70)\n",
            "epoch: 15710, dloss = 4.4, gloss = -2280.5, total_loss = -225.9 (0.68)\n",
            "epoch: 15711, dloss = 5.1, gloss = -2271.2, total_loss = -224.6 (0.68)\n",
            "epoch: 15712, dloss = -5.8, gloss = -2305.4, total_loss = -233.5 (0.74)\n",
            "epoch: 15713, dloss = 0.4, gloss = -2208.0, total_loss = -220.6 (1.00)\n",
            "epoch: 15714, dloss = 9.3, gloss = -2221.3, total_loss = -217.5 (0.68)\n",
            "epoch: 15715, dloss = 3.3, gloss = -2247.5, total_loss = -223.1 (0.68)\n",
            "epoch: 15716, dloss = 3.2, gloss = -2138.3, total_loss = -212.3 (0.68)\n",
            "epoch: 15717, dloss = 5.1, gloss = -2227.5, total_loss = -220.2 (0.69)\n",
            "epoch: 15718, dloss = 14.0, gloss = -2241.5, total_loss = -217.1 (0.67)\n",
            "epoch: 15719, dloss = 0.7, gloss = -2268.7, total_loss = -226.5 (0.69)\n",
            "epoch: 15720, dloss = 1.3, gloss = -2201.1, total_loss = -219.5 (0.70)\n",
            "epoch: 15721, dloss = -5.1, gloss = -2236.5, total_loss = -226.2 (0.68)\n",
            "epoch: 15722, dloss = -3.7, gloss = -2314.4, total_loss = -233.3 (0.68)\n",
            "epoch: 15723, dloss = -3.6, gloss = -2282.8, total_loss = -230.1 (0.69)\n",
            "epoch: 15724, dloss = 3.5, gloss = -2277.6, total_loss = -226.0 (0.66)\n",
            "epoch: 15725, dloss = 1.3, gloss = -2251.7, total_loss = -224.5 (0.67)\n",
            "epoch: 15726, dloss = 5.3, gloss = -2281.4, total_loss = -225.5 (0.70)\n",
            "epoch: 15727, dloss = 4.7, gloss = -2241.1, total_loss = -221.8 (0.67)\n",
            "epoch: 15728, dloss = 0.3, gloss = -2227.8, total_loss = -222.7 (0.85)\n",
            "epoch: 15729, dloss = -1.9, gloss = -2211.5, total_loss = -222.1 (0.85)\n",
            "epoch: 15730, dloss = -9.8, gloss = -2326.7, total_loss = -237.6 (0.70)\n",
            "epoch: 15731, dloss = 20.4, gloss = -2245.2, total_loss = -214.3 (0.69)\n",
            "epoch: 15732, dloss = 4.2, gloss = -2281.9, total_loss = -226.1 (0.67)\n",
            "epoch: 15733, dloss = 2.7, gloss = -2248.3, total_loss = -223.5 (0.69)\n",
            "epoch: 15734, dloss = -1.1, gloss = -2219.3, total_loss = -222.5 (0.68)\n",
            "epoch: 15735, dloss = 11.5, gloss = -2244.8, total_loss = -218.7 (0.68)\n",
            "epoch: 15736, dloss = 2.4, gloss = -2305.7, total_loss = -229.4 (0.69)\n",
            "epoch: 15737, dloss = 4.7, gloss = -2359.4, total_loss = -233.6 (0.69)\n",
            "epoch: 15738, dloss = -2.3, gloss = -2298.0, total_loss = -231.0 (0.69)\n",
            "epoch: 15739, dloss = 3.6, gloss = -2283.4, total_loss = -226.5 (0.67)\n",
            "epoch: 15740, dloss = 2.6, gloss = -2363.4, total_loss = -235.0 (0.68)\n",
            "epoch: 15741, dloss = 1.8, gloss = -2288.3, total_loss = -227.9 (0.70)\n",
            "epoch: 15742, dloss = 2.4, gloss = -2178.4, total_loss = -216.7 (0.70)\n",
            "epoch: 15743, dloss = 3.3, gloss = -2222.9, total_loss = -220.7 (0.69)\n",
            "epoch: 15744, dloss = 3.3, gloss = -2247.1, total_loss = -223.1 (1.00)\n",
            "epoch: 15745, dloss = -0.1, gloss = -2313.2, total_loss = -231.4 (0.71)\n",
            "epoch: 15746, dloss = 0.8, gloss = -2237.2, total_loss = -223.3 (0.69)\n",
            "epoch: 15747, dloss = 6.8, gloss = -2276.3, total_loss = -224.2 (0.66)\n",
            "epoch: 15748, dloss = -1.9, gloss = -2315.5, total_loss = -232.5 (0.68)\n",
            "epoch: 15749, dloss = 2.0, gloss = -2319.5, total_loss = -231.0 (0.67)\n",
            "epoch: 15750, dloss = 0.4, gloss = -2273.4, total_loss = -227.1 (0.69)\n",
            "epoch: 15751, dloss = 3.9, gloss = -2192.6, total_loss = -217.3 (0.69)\n",
            "epoch: 15752, dloss = -0.6, gloss = -2191.7, total_loss = -219.5 (0.68)\n",
            "epoch: 15753, dloss = 1.7, gloss = -2280.3, total_loss = -227.2 (0.67)\n",
            "epoch: 15754, dloss = 18.1, gloss = -2315.2, total_loss = -222.5 (0.68)\n",
            "epoch: 15755, dloss = 3.2, gloss = -2284.1, total_loss = -226.8 (0.68)\n",
            "epoch: 15756, dloss = 4.2, gloss = -2240.7, total_loss = -222.0 (0.68)\n",
            "epoch: 15757, dloss = -0.5, gloss = -2193.1, total_loss = -219.6 (0.69)\n",
            "epoch: 15758, dloss = -0.5, gloss = -2260.6, total_loss = -226.3 (0.69)\n",
            "epoch: 15759, dloss = 4.8, gloss = -2281.3, total_loss = -225.7 (0.72)\n",
            "epoch: 15760, dloss = 0.8, gloss = -2308.0, total_loss = -230.4 (0.99)\n",
            "epoch: 15761, dloss = -3.3, gloss = -2317.1, total_loss = -233.3 (0.69)\n",
            "epoch: 15762, dloss = 0.5, gloss = -2338.9, total_loss = -233.6 (0.70)\n",
            "epoch: 15763, dloss = 1.6, gloss = -2299.4, total_loss = -229.2 (0.67)\n",
            "epoch: 15764, dloss = 5.5, gloss = -2342.3, total_loss = -231.5 (0.68)\n",
            "epoch: 15765, dloss = 10.3, gloss = -2393.1, total_loss = -234.2 (0.68)\n",
            "epoch: 15766, dloss = -1.1, gloss = -2377.8, total_loss = -238.3 (0.69)\n",
            "epoch: 15767, dloss = -1.8, gloss = -2383.0, total_loss = -239.2 (0.70)\n",
            "epoch: 15768, dloss = -2.8, gloss = -2362.5, total_loss = -237.6 (0.68)\n",
            "epoch: 15769, dloss = 0.1, gloss = -2440.7, total_loss = -244.0 (0.68)\n",
            "epoch: 15770, dloss = 1.9, gloss = -2382.3, total_loss = -237.3 (0.68)\n",
            "epoch: 15771, dloss = -0.6, gloss = -2306.2, total_loss = -230.9 (0.67)\n",
            "epoch: 15772, dloss = 0.1, gloss = -2275.4, total_loss = -227.5 (0.68)\n",
            "epoch: 15773, dloss = -2.6, gloss = -2325.7, total_loss = -233.9 (0.69)\n",
            "epoch: 15774, dloss = 0.2, gloss = -2299.8, total_loss = -229.9 (0.67)\n",
            "epoch: 15775, dloss = 3.4, gloss = -2273.7, total_loss = -225.7 (0.85)\n",
            "epoch: 15776, dloss = -1.1, gloss = -2270.1, total_loss = -227.5 (0.86)\n",
            "epoch: 15777, dloss = 0.6, gloss = -2378.4, total_loss = -237.5 (0.70)\n",
            "epoch: 15778, dloss = -1.2, gloss = -2320.1, total_loss = -232.6 (0.69)\n",
            "epoch: 15779, dloss = -1.4, gloss = -2248.8, total_loss = -225.6 (0.69)\n",
            "epoch: 15780, dloss = -2.0, gloss = -2182.6, total_loss = -219.3 (0.70)\n",
            "epoch: 15781, dloss = -0.9, gloss = -2264.4, total_loss = -226.9 (0.68)\n",
            "epoch: 15782, dloss = -0.2, gloss = -2263.2, total_loss = -226.4 (0.69)\n",
            "epoch: 15783, dloss = -4.1, gloss = -2319.0, total_loss = -233.9 (0.69)\n",
            "epoch: 15784, dloss = -2.7, gloss = -2285.9, total_loss = -229.9 (0.69)\n",
            "epoch: 15785, dloss = 0.1, gloss = -2303.3, total_loss = -230.3 (0.69)\n",
            "epoch: 15786, dloss = 0.6, gloss = -2259.5, total_loss = -225.7 (0.69)\n",
            "epoch: 15787, dloss = 1.7, gloss = -2188.2, total_loss = -218.0 (0.69)\n",
            "epoch: 15788, dloss = 2.6, gloss = -2187.5, total_loss = -217.4 (0.67)\n",
            "epoch: 15789, dloss = 3.0, gloss = -2239.6, total_loss = -222.5 (0.70)\n",
            "epoch: 15790, dloss = -0.5, gloss = -2249.9, total_loss = -225.2 (0.71)\n",
            "epoch: 15791, dloss = 3.1, gloss = -2281.1, total_loss = -226.6 (1.02)\n",
            "epoch: 15792, dloss = 1.4, gloss = -2356.6, total_loss = -235.0 (0.69)\n",
            "epoch: 15793, dloss = -2.3, gloss = -2390.1, total_loss = -240.2 (0.68)\n",
            "epoch: 15794, dloss = 2.3, gloss = -2326.0, total_loss = -231.4 (0.68)\n",
            "epoch: 15795, dloss = -0.8, gloss = -2254.0, total_loss = -225.8 (0.70)\n",
            "epoch: 15796, dloss = 4.0, gloss = -2365.0, total_loss = -234.5 (0.68)\n",
            "epoch: 15797, dloss = -0.1, gloss = -2320.0, total_loss = -232.1 (0.69)\n",
            "epoch: 15798, dloss = 4.6, gloss = -2343.5, total_loss = -232.1 (0.71)\n",
            "epoch: 15799, dloss = 3.7, gloss = -2326.8, total_loss = -230.8 (0.67)\n",
            "epoch: 15800, dloss = 4.3, gloss = -2384.9, total_loss = -236.3 (0.69)\n",
            "epoch: 15801, dloss = 0.6, gloss = -2402.6, total_loss = -240.0 (0.68)\n",
            "epoch: 15802, dloss = -4.8, gloss = -2344.7, total_loss = -236.9 (0.69)\n",
            "epoch: 15803, dloss = -5.1, gloss = -2317.0, total_loss = -234.3 (0.70)\n",
            "epoch: 15804, dloss = 2.8, gloss = -2321.5, total_loss = -230.8 (0.68)\n",
            "epoch: 15805, dloss = 0.7, gloss = -2334.9, total_loss = -233.1 (0.67)\n",
            "epoch: 15806, dloss = 1.2, gloss = -2315.7, total_loss = -230.9 (0.87)\n",
            "epoch: 15807, dloss = -0.6, gloss = -2330.5, total_loss = -233.4 (1.06)\n",
            "epoch: 15808, dloss = -0.5, gloss = -2290.9, total_loss = -229.4 (0.89)\n",
            "epoch: 15809, dloss = 0.1, gloss = -2335.0, total_loss = -233.4 (0.68)\n",
            "epoch: 15810, dloss = 2.2, gloss = -2393.2, total_loss = -238.2 (0.67)\n",
            "epoch: 15811, dloss = 0.7, gloss = -2405.4, total_loss = -240.2 (0.68)\n",
            "epoch: 15812, dloss = 4.1, gloss = -2227.5, total_loss = -220.7 (0.69)\n",
            "epoch: 15813, dloss = 0.4, gloss = -2362.8, total_loss = -236.1 (0.69)\n",
            "epoch: 15814, dloss = 1.0, gloss = -2365.9, total_loss = -236.1 (0.87)\n",
            "epoch: 15815, dloss = 5.6, gloss = -2371.8, total_loss = -234.4 (1.10)\n",
            "epoch: 15816, dloss = 0.0, gloss = -2370.3, total_loss = -237.0 (0.97)\n",
            "epoch: 15817, dloss = -3.3, gloss = -2315.1, total_loss = -233.1 (0.69)\n",
            "epoch: 15818, dloss = 1.2, gloss = -2341.0, total_loss = -233.5 (0.69)\n",
            "epoch: 15819, dloss = 0.1, gloss = -2223.6, total_loss = -222.3 (0.69)\n",
            "epoch: 15820, dloss = -10.8, gloss = -2271.9, total_loss = -232.6 (0.90)\n",
            "epoch: 15821, dloss = -0.0, gloss = -2254.4, total_loss = -225.5 (0.84)\n",
            "epoch: 15822, dloss = -3.2, gloss = -2175.6, total_loss = -219.2 (0.70)\n",
            "epoch: 15823, dloss = 0.8, gloss = -2289.9, total_loss = -228.6 (0.68)\n",
            "epoch: 15824, dloss = 1.1, gloss = -2263.5, total_loss = -225.8 (0.68)\n",
            "epoch: 15825, dloss = 2.1, gloss = -2337.5, total_loss = -232.7 (0.68)\n",
            "epoch: 15826, dloss = 1.8, gloss = -2394.6, total_loss = -238.6 (0.68)\n",
            "epoch: 15827, dloss = 1.0, gloss = -2331.0, total_loss = -232.6 (0.71)\n",
            "epoch: 15828, dloss = 5.5, gloss = -2303.9, total_loss = -227.6 (0.71)\n",
            "epoch: 15829, dloss = 3.2, gloss = -2248.5, total_loss = -223.2 (0.69)\n",
            "epoch: 15830, dloss = 3.5, gloss = -2281.1, total_loss = -226.3 (0.72)\n",
            "epoch: 15831, dloss = -2.8, gloss = -2326.3, total_loss = -234.0 (0.68)\n",
            "epoch: 15832, dloss = 2.8, gloss = -2366.6, total_loss = -235.2 (0.69)\n",
            "epoch: 15833, dloss = -0.2, gloss = -2272.1, total_loss = -227.3 (0.70)\n",
            "epoch: 15834, dloss = 6.5, gloss = -2234.9, total_loss = -220.2 (0.69)\n",
            "epoch: 15835, dloss = 0.3, gloss = -2319.9, total_loss = -231.8 (0.78)\n",
            "epoch: 15836, dloss = 0.0, gloss = -2308.4, total_loss = -230.8 (0.97)\n",
            "epoch: 15837, dloss = -2.8, gloss = -2362.3, total_loss = -237.6 (0.70)\n",
            "epoch: 15838, dloss = 11.2, gloss = -2298.6, total_loss = -224.3 (0.68)\n",
            "epoch: 15839, dloss = -0.0, gloss = -2349.0, total_loss = -234.9 (0.68)\n",
            "epoch: 15840, dloss = 1.8, gloss = -2282.3, total_loss = -227.3 (0.70)\n",
            "epoch: 15841, dloss = 0.8, gloss = -2270.5, total_loss = -226.7 (0.69)\n",
            "epoch: 15842, dloss = 3.0, gloss = -2356.1, total_loss = -234.1 (0.70)\n",
            "epoch: 15843, dloss = 2.2, gloss = -2199.5, total_loss = -218.9 (0.69)\n",
            "epoch: 15844, dloss = 1.3, gloss = -2181.0, total_loss = -217.4 (0.69)\n",
            "epoch: 15845, dloss = -0.7, gloss = -2319.4, total_loss = -232.3 (0.68)\n",
            "epoch: 15846, dloss = -2.7, gloss = -2285.6, total_loss = -229.9 (0.70)\n",
            "epoch: 15847, dloss = -7.4, gloss = -2251.5, total_loss = -228.9 (0.70)\n",
            "epoch: 15848, dloss = 0.5, gloss = -2365.7, total_loss = -236.3 (0.67)\n",
            "epoch: 15849, dloss = -4.9, gloss = -2366.0, total_loss = -239.0 (0.68)\n",
            "epoch: 15850, dloss = -1.7, gloss = -2270.1, total_loss = -227.8 (0.70)\n",
            "epoch: 15851, dloss = 0.2, gloss = -2343.2, total_loss = -234.2 (0.98)\n",
            "epoch: 15852, dloss = 2.4, gloss = -2276.3, total_loss = -226.4 (0.78)\n",
            "epoch: 15853, dloss = 0.3, gloss = -2300.9, total_loss = -230.0 (0.65)\n",
            "epoch: 15854, dloss = 0.0, gloss = -2413.4, total_loss = -241.3 (0.66)\n",
            "epoch: 15855, dloss = 2.6, gloss = -2341.5, total_loss = -232.8 (0.70)\n",
            "epoch: 15856, dloss = -1.0, gloss = -2280.8, total_loss = -228.6 (0.69)\n",
            "epoch: 15857, dloss = 1.7, gloss = -2281.2, total_loss = -227.3 (0.67)\n",
            "epoch: 15858, dloss = -0.3, gloss = -2326.2, total_loss = -232.8 (0.70)\n",
            "epoch: 15859, dloss = 1.2, gloss = -2407.8, total_loss = -240.2 (0.68)\n",
            "epoch: 15860, dloss = -0.0, gloss = -2305.3, total_loss = -230.5 (0.70)\n",
            "epoch: 15861, dloss = 0.2, gloss = -2296.7, total_loss = -229.6 (0.69)\n",
            "epoch: 15862, dloss = -0.5, gloss = -2330.0, total_loss = -233.3 (0.69)\n",
            "epoch: 15863, dloss = 1.4, gloss = -2300.7, total_loss = -229.4 (0.69)\n",
            "epoch: 15864, dloss = -2.1, gloss = -2226.2, total_loss = -223.7 (0.67)\n",
            "epoch: 15865, dloss = 3.4, gloss = -2376.9, total_loss = -236.0 (0.69)\n",
            "epoch: 15866, dloss = -2.5, gloss = -2329.7, total_loss = -234.2 (0.79)\n",
            "epoch: 15867, dloss = 0.0, gloss = -2349.9, total_loss = -235.0 (0.98)\n",
            "epoch: 15868, dloss = 2.6, gloss = -2311.4, total_loss = -229.9 (0.69)\n",
            "epoch: 15869, dloss = 0.1, gloss = -2262.4, total_loss = -226.2 (0.66)\n",
            "epoch: 15870, dloss = -2.2, gloss = -2301.2, total_loss = -231.2 (0.69)\n",
            "epoch: 15871, dloss = -1.1, gloss = -2245.5, total_loss = -225.1 (0.69)\n",
            "epoch: 15872, dloss = -9.7, gloss = -2288.8, total_loss = -233.7 (0.69)\n",
            "epoch: 15873, dloss = 1.3, gloss = -2316.5, total_loss = -231.0 (0.67)\n",
            "epoch: 15874, dloss = -0.2, gloss = -2263.6, total_loss = -226.5 (0.67)\n",
            "epoch: 15875, dloss = 1.4, gloss = -2167.1, total_loss = -216.0 (0.69)\n",
            "epoch: 15876, dloss = -0.8, gloss = -2293.1, total_loss = -229.7 (0.69)\n",
            "epoch: 15877, dloss = 1.6, gloss = -2334.2, total_loss = -232.6 (0.68)\n",
            "epoch: 15878, dloss = 4.6, gloss = -2297.3, total_loss = -227.4 (0.65)\n",
            "epoch: 15879, dloss = 0.6, gloss = -2272.7, total_loss = -227.0 (0.70)\n",
            "epoch: 15880, dloss = -4.2, gloss = -2292.8, total_loss = -231.4 (0.67)\n",
            "epoch: 15881, dloss = -2.4, gloss = -2246.9, total_loss = -225.9 (0.64)\n",
            "epoch: 15882, dloss = -1.3, gloss = -2257.2, total_loss = -226.4 (0.82)\n",
            "epoch: 15883, dloss = -0.8, gloss = -2270.1, total_loss = -227.4 (0.89)\n",
            "epoch: 15884, dloss = 1.6, gloss = -2349.9, total_loss = -234.2 (0.69)\n",
            "epoch: 15885, dloss = 2.2, gloss = -2238.1, total_loss = -222.7 (0.67)\n",
            "epoch: 15886, dloss = 1.4, gloss = -2230.4, total_loss = -222.3 (0.70)\n",
            "epoch: 15887, dloss = 1.1, gloss = -2205.4, total_loss = -220.0 (0.70)\n",
            "epoch: 15888, dloss = 1.5, gloss = -2272.2, total_loss = -226.5 (0.68)\n",
            "epoch: 15889, dloss = -0.7, gloss = -2360.8, total_loss = -236.4 (0.70)\n",
            "epoch: 15890, dloss = 5.5, gloss = -2341.7, total_loss = -231.4 (0.68)\n",
            "epoch: 15891, dloss = -5.3, gloss = -2297.6, total_loss = -232.4 (0.67)\n",
            "epoch: 15892, dloss = -1.0, gloss = -2285.4, total_loss = -229.0 (0.67)\n",
            "epoch: 15893, dloss = 0.6, gloss = -2329.4, total_loss = -232.6 (0.67)\n",
            "epoch: 15894, dloss = 1.2, gloss = -2308.3, total_loss = -230.2 (0.68)\n",
            "epoch: 15895, dloss = -4.0, gloss = -2276.7, total_loss = -229.7 (0.69)\n",
            "epoch: 15896, dloss = 4.1, gloss = -2308.5, total_loss = -228.8 (0.70)\n",
            "epoch: 15897, dloss = -3.3, gloss = -2388.6, total_loss = -240.5 (0.68)\n",
            "epoch: 15898, dloss = -1.6, gloss = -2334.1, total_loss = -234.2 (0.98)\n",
            "epoch: 15899, dloss = 3.0, gloss = -2337.5, total_loss = -232.3 (0.75)\n",
            "epoch: 15900, dloss = -0.9, gloss = -2263.3, total_loss = -226.8 (0.67)\n",
            "epoch: 15901, dloss = 2.4, gloss = -2281.7, total_loss = -227.0 (0.67)\n",
            "epoch: 15902, dloss = -3.6, gloss = -2270.3, total_loss = -228.8 (0.70)\n",
            "epoch: 15903, dloss = -0.8, gloss = -2345.3, total_loss = -234.9 (0.70)\n",
            "epoch: 15904, dloss = 2.5, gloss = -2388.1, total_loss = -237.6 (0.70)\n",
            "epoch: 15905, dloss = -3.2, gloss = -2301.0, total_loss = -231.7 (0.67)\n",
            "epoch: 15906, dloss = -1.4, gloss = -2370.1, total_loss = -237.7 (0.70)\n",
            "epoch: 15907, dloss = -0.5, gloss = -2233.7, total_loss = -223.6 (0.67)\n",
            "epoch: 15908, dloss = -3.0, gloss = -2344.9, total_loss = -236.0 (0.66)\n",
            "epoch: 15909, dloss = 0.4, gloss = -2350.6, total_loss = -234.8 (0.70)\n",
            "epoch: 15910, dloss = 11.2, gloss = -2340.9, total_loss = -228.5 (0.68)\n",
            "epoch: 15911, dloss = -3.2, gloss = -2389.5, total_loss = -240.6 (0.67)\n",
            "epoch: 15912, dloss = 12.4, gloss = -2458.5, total_loss = -239.6 (0.69)\n",
            "epoch: 15913, dloss = 5.1, gloss = -2415.9, total_loss = -239.1 (0.75)\n",
            "epoch: 15914, dloss = 6.2, gloss = -2467.2, total_loss = -243.6 (0.96)\n",
            "epoch: 15915, dloss = 5.6, gloss = -2405.3, total_loss = -237.8 (0.67)\n",
            "epoch: 15916, dloss = 7.1, gloss = -2405.4, total_loss = -237.0 (0.66)\n",
            "epoch: 15917, dloss = 5.4, gloss = -2473.6, total_loss = -244.7 (0.68)\n",
            "epoch: 15918, dloss = 8.0, gloss = -2417.9, total_loss = -237.8 (0.69)\n",
            "epoch: 15919, dloss = -5.4, gloss = -2429.4, total_loss = -245.7 (0.66)\n",
            "epoch: 15920, dloss = 2.2, gloss = -2415.6, total_loss = -240.5 (0.69)\n",
            "epoch: 15921, dloss = 4.3, gloss = -2461.0, total_loss = -244.0 (0.68)\n",
            "epoch: 15922, dloss = 5.8, gloss = -2460.5, total_loss = -243.2 (0.68)\n",
            "epoch: 15923, dloss = 6.8, gloss = -2431.3, total_loss = -239.7 (0.67)\n",
            "epoch: 15924, dloss = 8.0, gloss = -2404.2, total_loss = -236.4 (0.67)\n",
            "epoch: 15925, dloss = 5.8, gloss = -2388.2, total_loss = -235.9 (0.68)\n",
            "epoch: 15926, dloss = 0.6, gloss = -2529.6, total_loss = -252.7 (0.69)\n",
            "epoch: 15927, dloss = 5.4, gloss = -2662.7, total_loss = -263.5 (0.66)\n",
            "epoch: 15928, dloss = 1.9, gloss = -2596.4, total_loss = -258.7 (0.69)\n",
            "epoch: 15929, dloss = 3.5, gloss = -2606.6, total_loss = -258.9 (0.85)\n",
            "epoch: 15930, dloss = 4.7, gloss = -2551.9, total_loss = -252.8 (0.87)\n",
            "epoch: 15931, dloss = 3.5, gloss = -2450.5, total_loss = -243.3 (0.69)\n",
            "epoch: 15932, dloss = 4.2, gloss = -2415.3, total_loss = -239.4 (0.69)\n",
            "epoch: 15933, dloss = -0.1, gloss = -2351.7, total_loss = -235.2 (0.69)\n",
            "epoch: 15934, dloss = 2.2, gloss = -2378.1, total_loss = -236.7 (0.67)\n",
            "epoch: 15935, dloss = 0.5, gloss = -2419.8, total_loss = -241.7 (0.68)\n",
            "epoch: 15936, dloss = 3.7, gloss = -2360.5, total_loss = -234.2 (0.66)\n",
            "epoch: 15937, dloss = 2.4, gloss = -2388.9, total_loss = -237.7 (0.68)\n",
            "epoch: 15938, dloss = 1.1, gloss = -2329.7, total_loss = -232.4 (0.68)\n",
            "epoch: 15939, dloss = -1.8, gloss = -2280.2, total_loss = -228.9 (0.67)\n",
            "epoch: 15940, dloss = -4.2, gloss = -2375.1, total_loss = -239.6 (0.69)\n",
            "epoch: 15941, dloss = -0.5, gloss = -2397.7, total_loss = -240.0 (0.68)\n",
            "epoch: 15942, dloss = -0.2, gloss = -2430.6, total_loss = -243.1 (0.68)\n",
            "epoch: 15943, dloss = -1.0, gloss = -2406.7, total_loss = -241.2 (0.67)\n",
            "epoch: 15944, dloss = -0.6, gloss = -2449.1, total_loss = -245.2 (0.68)\n",
            "epoch: 15945, dloss = 4.4, gloss = -2413.6, total_loss = -239.1 (0.92)\n",
            "epoch: 15946, dloss = 3.6, gloss = -2410.3, total_loss = -239.2 (0.80)\n",
            "epoch: 15947, dloss = 1.1, gloss = -2472.6, total_loss = -246.7 (0.69)\n",
            "epoch: 15948, dloss = 1.3, gloss = -2446.2, total_loss = -244.0 (0.69)\n",
            "epoch: 15949, dloss = -0.6, gloss = -2403.2, total_loss = -240.6 (0.70)\n",
            "epoch: 15950, dloss = -0.5, gloss = -2417.1, total_loss = -241.9 (0.69)\n",
            "epoch: 15951, dloss = -0.3, gloss = -2365.9, total_loss = -236.7 (0.70)\n",
            "epoch: 15952, dloss = -0.3, gloss = -2338.9, total_loss = -234.0 (0.67)\n",
            "epoch: 15953, dloss = 1.4, gloss = -2365.5, total_loss = -235.8 (0.69)\n",
            "epoch: 15954, dloss = 0.8, gloss = -2455.3, total_loss = -245.1 (0.70)\n",
            "epoch: 15955, dloss = 1.0, gloss = -2398.1, total_loss = -239.3 (0.69)\n",
            "epoch: 15956, dloss = 1.6, gloss = -2344.8, total_loss = -233.7 (0.68)\n",
            "epoch: 15957, dloss = 2.3, gloss = -2428.7, total_loss = -241.7 (0.70)\n",
            "epoch: 15958, dloss = 1.6, gloss = -2348.9, total_loss = -234.1 (0.71)\n",
            "epoch: 15959, dloss = -2.5, gloss = -2449.3, total_loss = -246.2 (0.68)\n",
            "epoch: 15960, dloss = 1.5, gloss = -2445.7, total_loss = -243.8 (0.78)\n",
            "epoch: 15961, dloss = -0.4, gloss = -2285.5, total_loss = -228.7 (0.98)\n",
            "epoch: 15962, dloss = 1.1, gloss = -2335.2, total_loss = -233.0 (0.71)\n",
            "epoch: 15963, dloss = 2.0, gloss = -2304.6, total_loss = -229.5 (0.67)\n",
            "epoch: 15964, dloss = 2.1, gloss = -2498.0, total_loss = -248.7 (0.69)\n",
            "epoch: 15965, dloss = 2.3, gloss = -2442.9, total_loss = -243.1 (0.68)\n",
            "epoch: 15966, dloss = 2.7, gloss = -2363.9, total_loss = -235.0 (0.67)\n",
            "epoch: 15967, dloss = -0.1, gloss = -2419.8, total_loss = -242.0 (0.68)\n",
            "epoch: 15968, dloss = -1.5, gloss = -2422.2, total_loss = -243.0 (0.68)\n",
            "epoch: 15969, dloss = -0.3, gloss = -2418.8, total_loss = -242.0 (0.69)\n",
            "epoch: 15970, dloss = 4.5, gloss = -2514.1, total_loss = -249.1 (0.68)\n",
            "epoch: 15971, dloss = -1.9, gloss = -2522.7, total_loss = -253.2 (0.71)\n",
            "epoch: 15972, dloss = 1.3, gloss = -2482.5, total_loss = -247.6 (0.70)\n",
            "epoch: 15973, dloss = 0.2, gloss = -2324.0, total_loss = -232.3 (0.67)\n",
            "epoch: 15974, dloss = 0.9, gloss = -2326.2, total_loss = -232.2 (0.71)\n",
            "epoch: 15975, dloss = 1.7, gloss = -2426.9, total_loss = -241.8 (0.69)\n",
            "epoch: 15976, dloss = -0.4, gloss = -2373.8, total_loss = -237.6 (0.95)\n",
            "epoch: 15977, dloss = 2.1, gloss = -2331.2, total_loss = -232.1 (0.79)\n",
            "epoch: 15978, dloss = -0.4, gloss = -2327.6, total_loss = -233.0 (0.71)\n",
            "epoch: 15979, dloss = 0.3, gloss = -2332.9, total_loss = -233.1 (0.70)\n",
            "epoch: 15980, dloss = 1.0, gloss = -2329.4, total_loss = -232.4 (0.68)\n",
            "epoch: 15981, dloss = -1.1, gloss = -2419.1, total_loss = -242.5 (0.72)\n",
            "epoch: 15982, dloss = -3.6, gloss = -2400.0, total_loss = -241.8 (0.69)\n",
            "epoch: 15983, dloss = 0.3, gloss = -2352.5, total_loss = -235.1 (0.70)\n",
            "epoch: 15984, dloss = 0.7, gloss = -2347.1, total_loss = -234.3 (0.68)\n",
            "epoch: 15985, dloss = -2.7, gloss = -2376.2, total_loss = -239.0 (0.70)\n",
            "epoch: 15986, dloss = -1.6, gloss = -2321.7, total_loss = -233.0 (0.71)\n",
            "epoch: 15987, dloss = 3.5, gloss = -2418.5, total_loss = -240.1 (0.69)\n",
            "epoch: 15988, dloss = 0.5, gloss = -2435.7, total_loss = -243.3 (0.72)\n",
            "epoch: 15989, dloss = 3.5, gloss = -2335.3, total_loss = -231.8 (0.68)\n",
            "epoch: 15990, dloss = -1.1, gloss = -2308.3, total_loss = -231.4 (0.73)\n",
            "epoch: 15991, dloss = 2.4, gloss = -2376.9, total_loss = -236.5 (0.87)\n",
            "epoch: 15992, dloss = 2.5, gloss = -2292.3, total_loss = -228.0 (0.88)\n",
            "epoch: 15993, dloss = -0.3, gloss = -2473.8, total_loss = -247.5 (0.70)\n",
            "epoch: 15994, dloss = 3.2, gloss = -2286.5, total_loss = -227.0 (0.70)\n",
            "epoch: 15995, dloss = 6.0, gloss = -2321.4, total_loss = -229.1 (0.71)\n",
            "epoch: 15996, dloss = 4.4, gloss = -2350.4, total_loss = -232.8 (0.69)\n",
            "epoch: 15997, dloss = 1.0, gloss = -2302.5, total_loss = -229.7 (0.68)\n",
            "epoch: 15998, dloss = 2.0, gloss = -2327.1, total_loss = -231.7 (0.70)\n",
            "epoch: 15999, dloss = 3.5, gloss = -2314.2, total_loss = -229.7 (0.69)\n",
            "epoch: 16000, dloss = 1.9, gloss = -2339.9, total_loss = -233.0 (0.69)\n",
            "epoch: 16001, dloss = 2.5, gloss = -2309.7, total_loss = -229.7 (0.68)\n",
            "epoch: 16002, dloss = 0.6, gloss = -2238.2, total_loss = -223.5 (0.69)\n",
            "epoch: 16003, dloss = 2.7, gloss = -2290.3, total_loss = -227.7 (0.67)\n",
            "epoch: 16004, dloss = 1.1, gloss = -2364.9, total_loss = -236.0 (0.69)\n",
            "epoch: 16005, dloss = 2.6, gloss = -2242.5, total_loss = -223.0 (0.69)\n",
            "epoch: 16006, dloss = 2.5, gloss = -2375.2, total_loss = -236.3 (0.75)\n",
            "epoch: 16007, dloss = 0.3, gloss = -2306.5, total_loss = -230.5 (1.01)\n",
            "epoch: 16008, dloss = -1.1, gloss = -2339.0, total_loss = -234.4 (0.70)\n",
            "epoch: 16009, dloss = -1.4, gloss = -2301.3, total_loss = -230.8 (0.70)\n",
            "epoch: 16010, dloss = -4.3, gloss = -2192.1, total_loss = -221.4 (0.69)\n",
            "epoch: 16011, dloss = 6.6, gloss = -2215.2, total_loss = -218.2 (0.70)\n",
            "epoch: 16012, dloss = 0.5, gloss = -2326.4, total_loss = -232.4 (0.68)\n",
            "epoch: 16013, dloss = 1.8, gloss = -2410.5, total_loss = -240.1 (0.68)\n",
            "epoch: 16014, dloss = -0.4, gloss = -2395.5, total_loss = -239.7 (0.70)\n",
            "epoch: 16015, dloss = -4.6, gloss = -2381.4, total_loss = -240.5 (0.69)\n",
            "epoch: 16016, dloss = 2.1, gloss = -2407.3, total_loss = -239.7 (0.70)\n",
            "epoch: 16017, dloss = -4.4, gloss = -2305.7, total_loss = -232.8 (0.67)\n",
            "epoch: 16018, dloss = 3.1, gloss = -2331.2, total_loss = -231.5 (0.71)\n",
            "epoch: 16019, dloss = -1.2, gloss = -2357.9, total_loss = -236.4 (0.70)\n",
            "epoch: 16020, dloss = 1.3, gloss = -2352.7, total_loss = -234.6 (0.68)\n",
            "epoch: 16021, dloss = 2.9, gloss = -2293.5, total_loss = -227.9 (0.70)\n",
            "epoch: 16022, dloss = -1.8, gloss = -2288.8, total_loss = -229.8 (0.94)\n",
            "epoch: 16023, dloss = 2.1, gloss = -2364.2, total_loss = -235.4 (0.83)\n",
            "epoch: 16024, dloss = -1.4, gloss = -2271.6, total_loss = -227.8 (0.69)\n",
            "epoch: 16025, dloss = 2.0, gloss = -2331.3, total_loss = -232.1 (0.70)\n",
            "epoch: 16026, dloss = -4.3, gloss = -2245.8, total_loss = -226.7 (0.68)\n",
            "epoch: 16027, dloss = 1.7, gloss = -2309.1, total_loss = -230.1 (0.68)\n",
            "epoch: 16028, dloss = -0.4, gloss = -2342.6, total_loss = -234.5 (0.69)\n",
            "epoch: 16029, dloss = 4.1, gloss = -2267.3, total_loss = -224.7 (0.69)\n",
            "epoch: 16030, dloss = 5.1, gloss = -2194.0, total_loss = -216.9 (0.67)\n",
            "epoch: 16031, dloss = 0.9, gloss = -2323.9, total_loss = -231.9 (0.67)\n",
            "epoch: 16032, dloss = -2.3, gloss = -2251.2, total_loss = -226.3 (0.70)\n",
            "epoch: 16033, dloss = 1.2, gloss = -2323.1, total_loss = -231.7 (0.69)\n",
            "epoch: 16034, dloss = -0.5, gloss = -2294.7, total_loss = -229.7 (0.69)\n",
            "epoch: 16035, dloss = -0.9, gloss = -2283.3, total_loss = -228.8 (0.70)\n",
            "epoch: 16036, dloss = -2.6, gloss = -2313.5, total_loss = -232.6 (0.66)\n",
            "epoch: 16037, dloss = 1.5, gloss = -2317.6, total_loss = -231.0 (0.73)\n",
            "epoch: 16038, dloss = 2.3, gloss = -2349.3, total_loss = -233.8 (1.01)\n",
            "epoch: 16039, dloss = 2.5, gloss = -2417.6, total_loss = -240.5 (0.69)\n",
            "epoch: 16040, dloss = 1.0, gloss = -2376.2, total_loss = -237.1 (0.69)\n",
            "epoch: 16041, dloss = 0.0, gloss = -2457.0, total_loss = -245.7 (0.70)\n",
            "epoch: 16042, dloss = -0.8, gloss = -2438.3, total_loss = -244.3 (0.68)\n",
            "epoch: 16043, dloss = 3.8, gloss = -2344.3, total_loss = -232.5 (0.68)\n",
            "epoch: 16044, dloss = 2.3, gloss = -2363.3, total_loss = -235.2 (0.68)\n",
            "epoch: 16045, dloss = -0.2, gloss = -2284.7, total_loss = -228.6 (0.68)\n",
            "epoch: 16046, dloss = 1.8, gloss = -2258.6, total_loss = -225.0 (0.68)\n",
            "epoch: 16047, dloss = -2.5, gloss = -2337.5, total_loss = -235.0 (0.67)\n",
            "epoch: 16048, dloss = -0.7, gloss = -2278.9, total_loss = -228.3 (0.68)\n",
            "epoch: 16049, dloss = -1.9, gloss = -2316.9, total_loss = -232.6 (0.67)\n",
            "epoch: 16050, dloss = 2.1, gloss = -2261.0, total_loss = -225.0 (0.65)\n",
            "epoch: 16051, dloss = 0.7, gloss = -2342.5, total_loss = -233.9 (0.69)\n",
            "epoch: 16052, dloss = -2.1, gloss = -2323.4, total_loss = -233.4 (0.67)\n",
            "epoch: 16053, dloss = 3.1, gloss = -2383.4, total_loss = -236.8 (0.86)\n",
            "epoch: 16054, dloss = 2.2, gloss = -2367.5, total_loss = -235.7 (0.88)\n",
            "epoch: 16055, dloss = 1.9, gloss = -2332.0, total_loss = -232.2 (0.68)\n",
            "epoch: 16056, dloss = 0.7, gloss = -2235.8, total_loss = -223.2 (0.68)\n",
            "epoch: 16057, dloss = -0.5, gloss = -2376.6, total_loss = -237.9 (0.67)\n",
            "epoch: 16058, dloss = -0.7, gloss = -2321.6, total_loss = -232.5 (0.76)\n",
            "epoch: 16059, dloss = 0.5, gloss = -2313.7, total_loss = -231.1 (0.97)\n",
            "epoch: 16060, dloss = 3.0, gloss = -2326.0, total_loss = -231.1 (0.71)\n",
            "epoch: 16061, dloss = -1.3, gloss = -2385.7, total_loss = -239.2 (0.67)\n",
            "epoch: 16062, dloss = -1.8, gloss = -2439.2, total_loss = -244.8 (0.72)\n",
            "epoch: 16063, dloss = -0.1, gloss = -2352.3, total_loss = -235.3 (0.68)\n",
            "epoch: 16064, dloss = 0.6, gloss = -2387.6, total_loss = -238.4 (0.66)\n",
            "epoch: 16065, dloss = 1.3, gloss = -2313.9, total_loss = -230.7 (0.69)\n",
            "epoch: 16066, dloss = -1.8, gloss = -2266.2, total_loss = -227.5 (0.70)\n",
            "epoch: 16067, dloss = -0.6, gloss = -2255.8, total_loss = -225.9 (0.68)\n",
            "epoch: 16068, dloss = -0.8, gloss = -2364.7, total_loss = -236.9 (1.14)\n",
            "epoch: 16069, dloss = 1.4, gloss = -2335.1, total_loss = -232.8 (1.02)\n",
            "epoch: 16070, dloss = -2.7, gloss = -2329.9, total_loss = -234.3 (0.68)\n",
            "epoch: 16071, dloss = -0.9, gloss = -2400.6, total_loss = -240.5 (0.68)\n",
            "epoch: 16072, dloss = 2.6, gloss = -2444.2, total_loss = -243.1 (0.67)\n",
            "epoch: 16073, dloss = 0.7, gloss = -2395.0, total_loss = -239.1 (0.67)\n",
            "epoch: 16074, dloss = 3.4, gloss = -2328.2, total_loss = -231.1 (0.69)\n",
            "epoch: 16075, dloss = -0.3, gloss = -2302.3, total_loss = -230.4 (0.68)\n",
            "epoch: 16076, dloss = -1.0, gloss = -2299.3, total_loss = -230.4 (0.67)\n",
            "epoch: 16077, dloss = 2.8, gloss = -2288.2, total_loss = -227.4 (0.69)\n",
            "epoch: 16078, dloss = -1.2, gloss = -2347.7, total_loss = -235.4 (0.67)\n",
            "epoch: 16079, dloss = -0.3, gloss = -2375.4, total_loss = -237.7 (0.68)\n",
            "epoch: 16080, dloss = -2.4, gloss = -2219.6, total_loss = -223.2 (0.69)\n",
            "epoch: 16081, dloss = -0.1, gloss = -2349.6, total_loss = -235.0 (0.68)\n",
            "epoch: 16082, dloss = 3.9, gloss = -2293.0, total_loss = -227.4 (0.69)\n",
            "epoch: 16083, dloss = 0.9, gloss = -2310.8, total_loss = -230.6 (0.67)\n",
            "epoch: 16084, dloss = 5.1, gloss = -2477.9, total_loss = -245.3 (0.86)\n",
            "epoch: 16085, dloss = 1.6, gloss = -2359.3, total_loss = -235.2 (0.87)\n",
            "epoch: 16086, dloss = -0.9, gloss = -2333.0, total_loss = -233.7 (0.69)\n",
            "epoch: 16087, dloss = 2.2, gloss = -2410.7, total_loss = -240.0 (0.68)\n",
            "epoch: 16088, dloss = 0.4, gloss = -2371.9, total_loss = -237.0 (0.66)\n",
            "epoch: 16089, dloss = 1.5, gloss = -2355.5, total_loss = -234.8 (0.67)\n",
            "epoch: 16090, dloss = 2.2, gloss = -2390.5, total_loss = -238.0 (0.68)\n",
            "epoch: 16091, dloss = 3.4, gloss = -2310.3, total_loss = -229.3 (0.68)\n",
            "epoch: 16092, dloss = 0.2, gloss = -2299.5, total_loss = -229.9 (0.70)\n",
            "epoch: 16093, dloss = -0.3, gloss = -2294.1, total_loss = -229.5 (0.68)\n",
            "epoch: 16094, dloss = 2.2, gloss = -2292.5, total_loss = -228.1 (0.68)\n",
            "epoch: 16095, dloss = 7.7, gloss = -2534.0, total_loss = -249.6 (0.67)\n",
            "epoch: 16096, dloss = 8.4, gloss = -2629.7, total_loss = -258.8 (0.69)\n",
            "epoch: 16097, dloss = 6.3, gloss = -2678.6, total_loss = -264.7 (0.68)\n",
            "epoch: 16098, dloss = 8.1, gloss = -2646.0, total_loss = -260.6 (0.68)\n",
            "epoch: 16099, dloss = 31.7, gloss = -2605.4, total_loss = -244.7 (0.68)\n",
            "epoch: 16100, dloss = 3.4, gloss = -2595.6, total_loss = -257.8 (0.98)\n",
            "epoch: 16101, dloss = 6.5, gloss = -2596.5, total_loss = -256.4 (0.77)\n",
            "epoch: 16102, dloss = 4.3, gloss = -2541.0, total_loss = -251.9 (0.67)\n",
            "epoch: 16103, dloss = 3.5, gloss = -2568.7, total_loss = -255.1 (0.69)\n",
            "epoch: 16104, dloss = 4.1, gloss = -2599.7, total_loss = -257.9 (0.67)\n",
            "epoch: 16105, dloss = 4.3, gloss = -2510.2, total_loss = -248.9 (0.69)\n",
            "epoch: 16106, dloss = 3.0, gloss = -2456.1, total_loss = -244.1 (0.68)\n",
            "epoch: 16107, dloss = 1.9, gloss = -2427.6, total_loss = -241.8 (0.70)\n",
            "epoch: 16108, dloss = 2.1, gloss = -2420.9, total_loss = -241.0 (0.97)\n",
            "epoch: 16109, dloss = 3.6, gloss = -2398.9, total_loss = -238.1 (0.86)\n",
            "epoch: 16110, dloss = 3.9, gloss = -2461.4, total_loss = -244.2 (0.68)\n",
            "epoch: 16111, dloss = 3.8, gloss = -2469.9, total_loss = -245.1 (0.67)\n",
            "epoch: 16112, dloss = 4.5, gloss = -2421.8, total_loss = -239.9 (0.67)\n",
            "epoch: 16113, dloss = 5.4, gloss = -2487.0, total_loss = -246.0 (0.68)\n",
            "epoch: 16114, dloss = 2.7, gloss = -2436.5, total_loss = -242.3 (0.67)\n",
            "epoch: 16115, dloss = 9.1, gloss = -2397.1, total_loss = -235.1 (0.96)\n",
            "epoch: 16116, dloss = 1.6, gloss = -2442.7, total_loss = -243.5 (0.76)\n",
            "epoch: 16117, dloss = 1.0, gloss = -2305.0, total_loss = -230.0 (0.70)\n",
            "epoch: 16118, dloss = 3.8, gloss = -2368.3, total_loss = -235.0 (0.71)\n",
            "epoch: 16119, dloss = 0.9, gloss = -2318.7, total_loss = -231.4 (0.69)\n",
            "epoch: 16120, dloss = 3.6, gloss = -2338.5, total_loss = -232.1 (0.67)\n",
            "epoch: 16121, dloss = 0.2, gloss = -2283.1, total_loss = -228.2 (0.66)\n",
            "epoch: 16122, dloss = -1.7, gloss = -2341.6, total_loss = -235.0 (0.68)\n",
            "epoch: 16123, dloss = -0.1, gloss = -2398.5, total_loss = -239.9 (0.69)\n",
            "epoch: 16124, dloss = 1.5, gloss = -2525.1, total_loss = -251.8 (0.68)\n",
            "epoch: 16125, dloss = 2.1, gloss = -2438.8, total_loss = -242.8 (0.66)\n",
            "epoch: 16126, dloss = -0.8, gloss = -2487.0, total_loss = -249.1 (0.68)\n",
            "epoch: 16127, dloss = 2.3, gloss = -2513.3, total_loss = -250.2 (0.68)\n",
            "epoch: 16128, dloss = 2.3, gloss = -2406.3, total_loss = -239.5 (0.67)\n",
            "epoch: 16129, dloss = 1.4, gloss = -2381.0, total_loss = -237.4 (0.68)\n",
            "epoch: 16130, dloss = -3.2, gloss = -2360.2, total_loss = -237.6 (0.69)\n",
            "epoch: 16131, dloss = 2.1, gloss = -2281.5, total_loss = -227.1 (1.01)\n",
            "epoch: 16132, dloss = 0.1, gloss = -2335.6, total_loss = -233.5 (0.66)\n",
            "epoch: 16133, dloss = -0.2, gloss = -2418.6, total_loss = -242.0 (0.67)\n",
            "epoch: 16134, dloss = 1.2, gloss = -2426.4, total_loss = -242.0 (0.70)\n",
            "epoch: 16135, dloss = -1.2, gloss = -2380.0, total_loss = -238.6 (0.69)\n",
            "epoch: 16136, dloss = 4.1, gloss = -2383.8, total_loss = -236.3 (0.67)\n",
            "epoch: 16137, dloss = 1.8, gloss = -2522.5, total_loss = -251.3 (0.66)\n",
            "epoch: 16138, dloss = 1.1, gloss = -2478.5, total_loss = -247.3 (0.67)\n",
            "epoch: 16139, dloss = 1.7, gloss = -2441.5, total_loss = -243.3 (0.68)\n",
            "epoch: 16140, dloss = -3.1, gloss = -2350.4, total_loss = -236.6 (0.69)\n",
            "epoch: 16141, dloss = 0.2, gloss = -2352.2, total_loss = -235.1 (0.67)\n",
            "epoch: 16142, dloss = 3.1, gloss = -2389.5, total_loss = -237.4 (0.70)\n",
            "epoch: 16143, dloss = 2.6, gloss = -2362.0, total_loss = -234.9 (0.68)\n",
            "epoch: 16144, dloss = 3.8, gloss = -2335.2, total_loss = -231.6 (0.68)\n",
            "epoch: 16145, dloss = 4.2, gloss = -2413.5, total_loss = -239.2 (0.69)\n",
            "epoch: 16146, dloss = -1.8, gloss = -2446.2, total_loss = -245.5 (0.78)\n",
            "epoch: 16147, dloss = 0.3, gloss = -2375.4, total_loss = -237.4 (0.92)\n",
            "epoch: 16148, dloss = -1.7, gloss = -2387.1, total_loss = -239.6 (0.68)\n",
            "epoch: 16149, dloss = 2.4, gloss = -2493.3, total_loss = -248.2 (0.71)\n",
            "epoch: 16150, dloss = 4.7, gloss = -2427.4, total_loss = -240.4 (0.69)\n",
            "epoch: 16151, dloss = 3.2, gloss = -2388.5, total_loss = -237.2 (0.68)\n",
            "epoch: 16152, dloss = 2.3, gloss = -2471.9, total_loss = -246.0 (0.69)\n",
            "epoch: 16153, dloss = 3.5, gloss = -2457.1, total_loss = -244.0 (0.69)\n",
            "epoch: 16154, dloss = 4.4, gloss = -2463.7, total_loss = -244.2 (0.69)\n",
            "epoch: 16155, dloss = 2.2, gloss = -2382.5, total_loss = -237.2 (0.70)\n",
            "epoch: 16156, dloss = 2.7, gloss = -2443.3, total_loss = -243.0 (0.72)\n",
            "epoch: 16157, dloss = 1.4, gloss = -2363.3, total_loss = -235.7 (0.69)\n",
            "epoch: 16158, dloss = -0.2, gloss = -2466.9, total_loss = -246.8 (0.69)\n",
            "epoch: 16159, dloss = -0.2, gloss = -2379.9, total_loss = -238.1 (0.69)\n",
            "epoch: 16160, dloss = -0.4, gloss = -2356.4, total_loss = -235.9 (0.67)\n",
            "epoch: 16161, dloss = 1.1, gloss = -2345.6, total_loss = -234.0 (0.70)\n",
            "epoch: 16162, dloss = 2.0, gloss = -2377.0, total_loss = -236.7 (1.01)\n",
            "epoch: 16163, dloss = 2.2, gloss = -2389.8, total_loss = -237.9 (0.72)\n",
            "epoch: 16164, dloss = 5.1, gloss = -2403.5, total_loss = -237.8 (0.68)\n",
            "epoch: 16165, dloss = -1.7, gloss = -2401.7, total_loss = -241.0 (0.69)\n",
            "epoch: 16166, dloss = 2.5, gloss = -2393.7, total_loss = -238.1 (0.68)\n",
            "epoch: 16167, dloss = 2.2, gloss = -2362.8, total_loss = -235.2 (0.68)\n",
            "epoch: 16168, dloss = 2.0, gloss = -2384.7, total_loss = -237.5 (0.69)\n",
            "epoch: 16169, dloss = 4.1, gloss = -2401.0, total_loss = -238.1 (0.68)\n",
            "epoch: 16170, dloss = 5.7, gloss = -2460.0, total_loss = -243.1 (0.68)\n",
            "epoch: 16171, dloss = 4.5, gloss = -2384.2, total_loss = -236.2 (0.69)\n",
            "epoch: 16172, dloss = 6.2, gloss = -2393.9, total_loss = -236.3 (0.68)\n",
            "epoch: 16173, dloss = -0.5, gloss = -2394.5, total_loss = -239.7 (0.68)\n",
            "epoch: 16174, dloss = -4.9, gloss = -2404.7, total_loss = -242.9 (0.70)\n",
            "epoch: 16175, dloss = -1.9, gloss = -2484.7, total_loss = -249.4 (0.69)\n",
            "epoch: 16176, dloss = 1.7, gloss = -2368.7, total_loss = -236.0 (0.68)\n",
            "epoch: 16177, dloss = -1.0, gloss = -2352.5, total_loss = -235.8 (0.78)\n",
            "epoch: 16178, dloss = -0.8, gloss = -2332.1, total_loss = -233.6 (0.91)\n",
            "epoch: 16179, dloss = 1.1, gloss = -2374.4, total_loss = -236.9 (0.68)\n",
            "epoch: 16180, dloss = 2.7, gloss = -2357.1, total_loss = -234.4 (0.67)\n",
            "epoch: 16181, dloss = 3.1, gloss = -2299.7, total_loss = -228.4 (0.70)\n",
            "epoch: 16182, dloss = -2.4, gloss = -2424.7, total_loss = -243.7 (0.69)\n",
            "epoch: 16183, dloss = -3.1, gloss = -2388.7, total_loss = -240.4 (0.66)\n",
            "epoch: 16184, dloss = 10.0, gloss = -2441.7, total_loss = -239.2 (0.69)\n",
            "epoch: 16185, dloss = 2.4, gloss = -2439.5, total_loss = -242.8 (0.69)\n",
            "epoch: 16186, dloss = 3.0, gloss = -2435.1, total_loss = -242.0 (0.70)\n",
            "epoch: 16187, dloss = 1.3, gloss = -2417.3, total_loss = -241.1 (0.68)\n",
            "epoch: 16188, dloss = -1.6, gloss = -2489.5, total_loss = -249.8 (0.69)\n",
            "epoch: 16189, dloss = 1.5, gloss = -2576.3, total_loss = -256.9 (0.69)\n",
            "epoch: 16190, dloss = 4.7, gloss = -2532.8, total_loss = -250.9 (0.67)\n",
            "epoch: 16191, dloss = -0.3, gloss = -2518.9, total_loss = -252.0 (0.68)\n",
            "epoch: 16192, dloss = 3.1, gloss = -2481.4, total_loss = -246.6 (0.71)\n",
            "epoch: 16193, dloss = 3.1, gloss = -2375.8, total_loss = -236.0 (0.94)\n",
            "epoch: 16194, dloss = 2.7, gloss = -2451.5, total_loss = -243.8 (0.80)\n",
            "epoch: 16195, dloss = 1.9, gloss = -2461.0, total_loss = -245.1 (0.67)\n",
            "epoch: 16196, dloss = -1.2, gloss = -2332.0, total_loss = -233.8 (0.69)\n",
            "epoch: 16197, dloss = 1.5, gloss = -2473.3, total_loss = -246.6 (0.65)\n",
            "epoch: 16198, dloss = 2.5, gloss = -2390.6, total_loss = -237.8 (0.67)\n",
            "epoch: 16199, dloss = 3.4, gloss = -2465.7, total_loss = -244.9 (0.69)\n",
            "epoch: 16200, dloss = -2.5, gloss = -2443.2, total_loss = -245.6 (0.67)\n",
            "epoch: 16201, dloss = 1.2, gloss = -2435.0, total_loss = -242.9 (0.67)\n",
            "epoch: 16202, dloss = 0.0, gloss = -2351.8, total_loss = -235.2 (0.67)\n",
            "epoch: 16203, dloss = 0.4, gloss = -2313.4, total_loss = -231.1 (0.69)\n",
            "epoch: 16204, dloss = -1.1, gloss = -2389.8, total_loss = -239.5 (0.70)\n",
            "epoch: 16205, dloss = 4.0, gloss = -2430.0, total_loss = -241.0 (0.68)\n",
            "epoch: 16206, dloss = 1.8, gloss = -2318.7, total_loss = -231.0 (0.67)\n",
            "epoch: 16207, dloss = 3.0, gloss = -2389.4, total_loss = -237.4 (0.67)\n",
            "epoch: 16208, dloss = 3.7, gloss = -2563.0, total_loss = -254.5 (0.68)\n",
            "epoch: 16209, dloss = -0.6, gloss = -2456.4, total_loss = -246.0 (1.06)\n",
            "epoch: 16210, dloss = 6.9, gloss = -2466.6, total_loss = -243.2 (0.68)\n",
            "epoch: 16211, dloss = 2.4, gloss = -2365.6, total_loss = -235.3 (0.69)\n",
            "epoch: 16212, dloss = 8.2, gloss = -2370.6, total_loss = -233.0 (0.69)\n",
            "epoch: 16213, dloss = 5.2, gloss = -2361.9, total_loss = -233.6 (0.67)\n",
            "epoch: 16214, dloss = -0.8, gloss = -2383.1, total_loss = -238.7 (0.69)\n",
            "epoch: 16215, dloss = 0.0, gloss = -2387.8, total_loss = -238.8 (0.67)\n",
            "epoch: 16216, dloss = 5.3, gloss = -2467.4, total_loss = -244.1 (0.67)\n",
            "epoch: 16217, dloss = 6.0, gloss = -2595.0, total_loss = -256.5 (0.69)\n",
            "epoch: 16218, dloss = 3.6, gloss = -2545.8, total_loss = -252.8 (0.68)\n",
            "epoch: 16219, dloss = 5.6, gloss = -2547.1, total_loss = -251.9 (0.68)\n",
            "epoch: 16220, dloss = 6.3, gloss = -2608.5, total_loss = -257.7 (0.69)\n",
            "epoch: 16221, dloss = 4.3, gloss = -2589.1, total_loss = -256.8 (0.68)\n",
            "epoch: 16222, dloss = 7.3, gloss = -2614.6, total_loss = -257.8 (0.67)\n",
            "epoch: 16223, dloss = 4.1, gloss = -2701.5, total_loss = -268.1 (0.69)\n",
            "epoch: 16224, dloss = 4.8, gloss = -2710.4, total_loss = -268.6 (0.78)\n",
            "epoch: 16225, dloss = 6.1, gloss = -2685.5, total_loss = -265.5 (0.94)\n",
            "epoch: 16226, dloss = 3.2, gloss = -2704.4, total_loss = -268.9 (0.69)\n",
            "epoch: 16227, dloss = 4.1, gloss = -2670.0, total_loss = -265.0 (0.67)\n",
            "epoch: 16228, dloss = -1.9, gloss = -2662.4, total_loss = -267.2 (0.68)\n",
            "epoch: 16229, dloss = 6.5, gloss = -2702.8, total_loss = -267.0 (0.67)\n",
            "epoch: 16230, dloss = 3.6, gloss = -2690.5, total_loss = -267.3 (0.70)\n",
            "epoch: 16231, dloss = 3.9, gloss = -2629.9, total_loss = -261.0 (0.68)\n",
            "epoch: 16232, dloss = 4.7, gloss = -2688.1, total_loss = -266.4 (0.67)\n",
            "epoch: 16233, dloss = 2.2, gloss = -2691.3, total_loss = -268.0 (0.69)\n",
            "epoch: 16234, dloss = 3.1, gloss = -2686.2, total_loss = -267.1 (0.69)\n",
            "epoch: 16235, dloss = 1.4, gloss = -2635.3, total_loss = -262.8 (0.70)\n",
            "epoch: 16236, dloss = -0.6, gloss = -2617.9, total_loss = -262.1 (0.67)\n",
            "epoch: 16237, dloss = -0.1, gloss = -2576.2, total_loss = -257.7 (0.69)\n",
            "epoch: 16238, dloss = 1.7, gloss = -2569.0, total_loss = -256.0 (0.68)\n",
            "epoch: 16239, dloss = 3.6, gloss = -2543.4, total_loss = -252.5 (0.70)\n",
            "epoch: 16240, dloss = 4.6, gloss = -2646.0, total_loss = -262.3 (0.92)\n",
            "epoch: 16241, dloss = 5.9, gloss = -2640.5, total_loss = -261.1 (0.82)\n",
            "epoch: 16242, dloss = 4.2, gloss = -2670.4, total_loss = -264.9 (0.68)\n",
            "epoch: 16243, dloss = 4.2, gloss = -2627.1, total_loss = -260.6 (0.69)\n",
            "epoch: 16244, dloss = 4.9, gloss = -2642.4, total_loss = -261.8 (0.67)\n",
            "epoch: 16245, dloss = 0.4, gloss = -2637.1, total_loss = -263.5 (0.68)\n",
            "epoch: 16246, dloss = 3.5, gloss = -2665.8, total_loss = -264.8 (0.69)\n",
            "epoch: 16247, dloss = 1.6, gloss = -2651.6, total_loss = -264.3 (0.68)\n",
            "epoch: 16248, dloss = 4.5, gloss = -2659.2, total_loss = -263.7 (0.68)\n",
            "epoch: 16249, dloss = 3.6, gloss = -2655.7, total_loss = -263.8 (0.68)\n",
            "epoch: 16250, dloss = 2.2, gloss = -2614.6, total_loss = -260.3 (0.67)\n",
            "epoch: 16251, dloss = 3.6, gloss = -2639.9, total_loss = -262.2 (0.67)\n",
            "epoch: 16252, dloss = 3.9, gloss = -2621.6, total_loss = -260.2 (0.68)\n",
            "epoch: 16253, dloss = 2.0, gloss = -2597.2, total_loss = -258.7 (0.67)\n",
            "epoch: 16254, dloss = 4.3, gloss = -2570.8, total_loss = -254.9 (0.68)\n",
            "epoch: 16255, dloss = 1.2, gloss = -2576.9, total_loss = -257.1 (1.26)\n",
            "epoch: 16256, dloss = -0.6, gloss = -2600.5, total_loss = -260.3 (0.69)\n",
            "epoch: 16257, dloss = -0.5, gloss = -2580.8, total_loss = -258.3 (0.69)\n",
            "epoch: 16258, dloss = 0.8, gloss = -2522.7, total_loss = -251.9 (0.71)\n",
            "epoch: 16259, dloss = 5.3, gloss = -2500.2, total_loss = -247.4 (0.70)\n",
            "epoch: 16260, dloss = 3.5, gloss = -2585.5, total_loss = -256.8 (0.69)\n",
            "epoch: 16261, dloss = -1.2, gloss = -2525.2, total_loss = -253.1 (0.70)\n",
            "epoch: 16262, dloss = 0.6, gloss = -2572.9, total_loss = -257.0 (0.69)\n",
            "epoch: 16263, dloss = 3.0, gloss = -2553.7, total_loss = -253.9 (0.71)\n",
            "epoch: 16264, dloss = 3.1, gloss = -2599.8, total_loss = -258.4 (0.67)\n",
            "epoch: 16265, dloss = 3.3, gloss = -2531.2, total_loss = -251.5 (0.68)\n",
            "epoch: 16266, dloss = 5.2, gloss = -2573.3, total_loss = -254.7 (0.70)\n",
            "epoch: 16267, dloss = 3.7, gloss = -2497.9, total_loss = -247.9 (0.69)\n",
            "epoch: 16268, dloss = 2.6, gloss = -2534.1, total_loss = -252.1 (0.66)\n",
            "epoch: 16269, dloss = 3.7, gloss = -2546.4, total_loss = -252.8 (0.70)\n",
            "epoch: 16270, dloss = 5.2, gloss = -2589.6, total_loss = -256.3 (0.84)\n",
            "epoch: 16271, dloss = -0.3, gloss = -2517.3, total_loss = -251.9 (0.87)\n",
            "epoch: 16272, dloss = 0.3, gloss = -2465.3, total_loss = -246.4 (0.69)\n",
            "epoch: 16273, dloss = -0.5, gloss = -2483.1, total_loss = -248.6 (0.70)\n",
            "epoch: 16274, dloss = 3.4, gloss = -2417.4, total_loss = -240.0 (0.67)\n",
            "epoch: 16275, dloss = 3.3, gloss = -2441.9, total_loss = -242.5 (0.69)\n",
            "epoch: 16276, dloss = 2.6, gloss = -2481.5, total_loss = -246.9 (0.67)\n",
            "epoch: 16277, dloss = 4.8, gloss = -2540.2, total_loss = -251.6 (0.68)\n",
            "epoch: 16278, dloss = 3.7, gloss = -2526.3, total_loss = -250.8 (0.69)\n",
            "epoch: 16279, dloss = 2.9, gloss = -2561.3, total_loss = -254.7 (0.65)\n",
            "epoch: 16280, dloss = 3.3, gloss = -2524.7, total_loss = -250.8 (0.69)\n",
            "epoch: 16281, dloss = 4.3, gloss = -2509.6, total_loss = -248.8 (0.68)\n",
            "epoch: 16282, dloss = 1.5, gloss = -2482.8, total_loss = -247.5 (0.70)\n",
            "epoch: 16283, dloss = 3.8, gloss = -2430.1, total_loss = -241.1 (0.68)\n",
            "epoch: 16284, dloss = 3.5, gloss = -2404.5, total_loss = -238.7 (0.69)\n",
            "epoch: 16285, dloss = -0.9, gloss = -2372.5, total_loss = -237.7 (0.69)\n",
            "epoch: 16286, dloss = 3.8, gloss = -2367.0, total_loss = -234.8 (0.97)\n",
            "epoch: 16287, dloss = 3.3, gloss = -2382.2, total_loss = -236.6 (0.76)\n",
            "epoch: 16288, dloss = -0.0, gloss = -2409.7, total_loss = -241.0 (0.70)\n",
            "epoch: 16289, dloss = 0.1, gloss = -2387.4, total_loss = -238.7 (0.68)\n",
            "epoch: 16290, dloss = -2.4, gloss = -2386.5, total_loss = -239.8 (0.72)\n",
            "epoch: 16291, dloss = 2.0, gloss = -2434.7, total_loss = -242.5 (0.69)\n",
            "epoch: 16292, dloss = 2.6, gloss = -2422.1, total_loss = -240.9 (0.67)\n",
            "epoch: 16293, dloss = 0.9, gloss = -2365.6, total_loss = -236.1 (0.68)\n",
            "epoch: 16294, dloss = 2.7, gloss = -2349.5, total_loss = -233.6 (0.68)\n",
            "epoch: 16295, dloss = -0.2, gloss = -2457.7, total_loss = -245.9 (0.67)\n",
            "epoch: 16296, dloss = 0.6, gloss = -2444.5, total_loss = -244.2 (0.71)\n",
            "epoch: 16297, dloss = 0.1, gloss = -2395.7, total_loss = -239.5 (0.70)\n",
            "epoch: 16298, dloss = 1.7, gloss = -2361.1, total_loss = -235.3 (0.70)\n",
            "epoch: 16299, dloss = 1.8, gloss = -2404.4, total_loss = -239.6 (0.70)\n",
            "epoch: 16300, dloss = 1.4, gloss = -2371.7, total_loss = -236.5 (0.72)\n",
            "epoch: 16301, dloss = 3.7, gloss = -2438.2, total_loss = -242.0 (0.86)\n",
            "epoch: 16302, dloss = 2.5, gloss = -2397.5, total_loss = -238.5 (0.88)\n",
            "epoch: 16303, dloss = 7.0, gloss = -2513.1, total_loss = -247.8 (0.70)\n",
            "epoch: 16304, dloss = 1.4, gloss = -2419.9, total_loss = -241.3 (0.70)\n",
            "epoch: 16305, dloss = 2.8, gloss = -2392.2, total_loss = -237.8 (0.70)\n",
            "epoch: 16306, dloss = 2.7, gloss = -2433.5, total_loss = -242.0 (0.72)\n",
            "epoch: 16307, dloss = 0.9, gloss = -2357.3, total_loss = -235.3 (0.69)\n",
            "epoch: 16308, dloss = 1.5, gloss = -2409.1, total_loss = -240.2 (0.70)\n",
            "epoch: 16309, dloss = -0.4, gloss = -2444.9, total_loss = -244.7 (0.68)\n",
            "epoch: 16310, dloss = 4.4, gloss = -2442.3, total_loss = -242.0 (0.71)\n",
            "epoch: 16311, dloss = 2.7, gloss = -2357.2, total_loss = -234.4 (1.08)\n",
            "epoch: 16312, dloss = 3.3, gloss = -2381.0, total_loss = -236.4 (0.70)\n",
            "epoch: 16313, dloss = 1.5, gloss = -2336.3, total_loss = -232.9 (0.69)\n",
            "epoch: 16314, dloss = -0.7, gloss = -2345.6, total_loss = -234.9 (0.70)\n",
            "epoch: 16315, dloss = 10.1, gloss = -2415.8, total_loss = -236.6 (0.68)\n",
            "epoch: 16316, dloss = 2.6, gloss = -2410.4, total_loss = -239.8 (0.95)\n",
            "epoch: 16317, dloss = 3.4, gloss = -2451.7, total_loss = -243.5 (0.81)\n",
            "epoch: 16318, dloss = 3.2, gloss = -2391.5, total_loss = -237.6 (0.71)\n",
            "epoch: 16319, dloss = 1.0, gloss = -2392.7, total_loss = -238.8 (0.69)\n",
            "epoch: 16320, dloss = 4.1, gloss = -2411.7, total_loss = -239.1 (0.69)\n",
            "epoch: 16321, dloss = 1.0, gloss = -2433.5, total_loss = -242.8 (0.94)\n",
            "epoch: 16322, dloss = 1.4, gloss = -2461.9, total_loss = -245.5 (0.80)\n",
            "epoch: 16323, dloss = 2.9, gloss = -2475.2, total_loss = -246.0 (0.71)\n",
            "epoch: 16324, dloss = 1.3, gloss = -2497.2, total_loss = -249.0 (0.69)\n",
            "epoch: 16325, dloss = 8.1, gloss = -2476.0, total_loss = -243.5 (0.68)\n",
            "epoch: 16326, dloss = 4.3, gloss = -2430.5, total_loss = -240.9 (0.71)\n",
            "epoch: 16327, dloss = 2.2, gloss = -2352.6, total_loss = -234.1 (0.70)\n",
            "epoch: 16328, dloss = 3.8, gloss = -2425.0, total_loss = -240.6 (0.70)\n",
            "epoch: 16329, dloss = 4.1, gloss = -2465.9, total_loss = -244.5 (0.72)\n",
            "epoch: 16330, dloss = 0.1, gloss = -2500.1, total_loss = -250.0 (0.71)\n",
            "epoch: 16331, dloss = 3.6, gloss = -2549.9, total_loss = -253.2 (1.02)\n",
            "epoch: 16332, dloss = 2.1, gloss = -2546.5, total_loss = -253.6 (0.71)\n",
            "epoch: 16333, dloss = 2.9, gloss = -2544.8, total_loss = -253.0 (0.69)\n",
            "epoch: 16334, dloss = 4.0, gloss = -2500.6, total_loss = -248.0 (0.71)\n",
            "epoch: 16335, dloss = 3.0, gloss = -2537.2, total_loss = -252.2 (0.68)\n",
            "epoch: 16336, dloss = 5.2, gloss = -2538.8, total_loss = -251.3 (0.70)\n",
            "epoch: 16337, dloss = 6.3, gloss = -2558.6, total_loss = -252.7 (0.70)\n",
            "epoch: 16338, dloss = 7.4, gloss = -2576.8, total_loss = -254.0 (0.69)\n",
            "epoch: 16339, dloss = 6.4, gloss = -2630.5, total_loss = -259.8 (0.70)\n",
            "epoch: 16340, dloss = 9.0, gloss = -2623.8, total_loss = -257.9 (0.68)\n",
            "epoch: 16341, dloss = 4.2, gloss = -2647.5, total_loss = -262.6 (0.68)\n",
            "epoch: 16342, dloss = 4.2, gloss = -2656.7, total_loss = -263.6 (0.69)\n",
            "epoch: 16343, dloss = 4.2, gloss = -2624.7, total_loss = -260.4 (0.68)\n",
            "epoch: 16344, dloss = 2.4, gloss = -2612.5, total_loss = -260.0 (0.67)\n",
            "epoch: 16345, dloss = 4.7, gloss = -2613.8, total_loss = -259.0 (0.69)\n",
            "epoch: 16346, dloss = 2.7, gloss = -2654.4, total_loss = -264.1 (0.84)\n",
            "epoch: 16347, dloss = 6.5, gloss = -2607.4, total_loss = -257.5 (0.90)\n",
            "epoch: 16348, dloss = 5.5, gloss = -2563.8, total_loss = -253.6 (0.67)\n",
            "epoch: 16349, dloss = 4.3, gloss = -2572.0, total_loss = -255.1 (0.69)\n",
            "epoch: 16350, dloss = 5.4, gloss = -2523.3, total_loss = -249.6 (0.68)\n",
            "epoch: 16351, dloss = 3.1, gloss = -2502.6, total_loss = -248.7 (0.65)\n",
            "epoch: 16352, dloss = 2.1, gloss = -2442.7, total_loss = -243.2 (0.69)\n",
            "epoch: 16353, dloss = 4.3, gloss = -2491.2, total_loss = -247.0 (0.71)\n",
            "epoch: 16354, dloss = 2.5, gloss = -2514.1, total_loss = -250.1 (0.70)\n",
            "epoch: 16355, dloss = 0.3, gloss = -2484.0, total_loss = -248.2 (0.69)\n",
            "epoch: 16356, dloss = -0.6, gloss = -2519.4, total_loss = -252.3 (0.68)\n",
            "epoch: 16357, dloss = 2.3, gloss = -2477.8, total_loss = -246.6 (0.67)\n",
            "epoch: 16358, dloss = 4.6, gloss = -2453.4, total_loss = -243.0 (0.69)\n",
            "epoch: 16359, dloss = 2.3, gloss = -2501.7, total_loss = -249.0 (0.69)\n",
            "epoch: 16360, dloss = 2.7, gloss = -2453.0, total_loss = -243.9 (0.69)\n",
            "epoch: 16361, dloss = 1.7, gloss = -2473.4, total_loss = -246.5 (0.66)\n",
            "epoch: 16362, dloss = 2.1, gloss = -2463.0, total_loss = -245.2 (0.98)\n",
            "epoch: 16363, dloss = 1.0, gloss = -2406.4, total_loss = -240.1 (0.73)\n",
            "epoch: 16364, dloss = 1.7, gloss = -2386.7, total_loss = -237.8 (0.69)\n",
            "epoch: 16365, dloss = 4.0, gloss = -2499.5, total_loss = -247.9 (0.68)\n",
            "epoch: 16366, dloss = 5.1, gloss = -2494.8, total_loss = -247.0 (0.66)\n",
            "epoch: 16367, dloss = 5.6, gloss = -2540.5, total_loss = -251.3 (0.71)\n",
            "epoch: 16368, dloss = 4.5, gloss = -2537.0, total_loss = -251.4 (0.69)\n",
            "epoch: 16369, dloss = 5.4, gloss = -2493.3, total_loss = -246.6 (0.71)\n",
            "epoch: 16370, dloss = 3.3, gloss = -2483.1, total_loss = -246.7 (0.70)\n",
            "epoch: 16371, dloss = 7.0, gloss = -2574.9, total_loss = -254.0 (0.70)\n",
            "epoch: 16372, dloss = 5.3, gloss = -2473.6, total_loss = -244.7 (0.69)\n",
            "epoch: 16373, dloss = 3.4, gloss = -2490.7, total_loss = -247.4 (0.69)\n",
            "epoch: 16374, dloss = 3.7, gloss = -2469.5, total_loss = -245.1 (0.68)\n",
            "epoch: 16375, dloss = 2.7, gloss = -2522.5, total_loss = -250.9 (0.70)\n",
            "epoch: 16376, dloss = -0.2, gloss = -2542.8, total_loss = -254.4 (0.69)\n",
            "epoch: 16377, dloss = 3.2, gloss = -2507.4, total_loss = -249.1 (0.83)\n",
            "epoch: 16378, dloss = 3.3, gloss = -2512.0, total_loss = -249.5 (0.86)\n",
            "epoch: 16379, dloss = -2.3, gloss = -2542.2, total_loss = -255.4 (0.67)\n",
            "epoch: 16380, dloss = 1.0, gloss = -2525.4, total_loss = -252.0 (0.68)\n",
            "epoch: 16381, dloss = 0.5, gloss = -2462.5, total_loss = -246.0 (0.69)\n",
            "epoch: 16382, dloss = 2.4, gloss = -2458.3, total_loss = -244.6 (0.71)\n",
            "epoch: 16383, dloss = 1.4, gloss = -2433.8, total_loss = -242.7 (0.69)\n",
            "epoch: 16384, dloss = 1.3, gloss = -2359.1, total_loss = -235.3 (0.67)\n",
            "epoch: 16385, dloss = -2.3, gloss = -2442.0, total_loss = -245.3 (0.67)\n",
            "epoch: 16386, dloss = 3.6, gloss = -2462.4, total_loss = -244.5 (0.68)\n",
            "epoch: 16387, dloss = 2.8, gloss = -2470.3, total_loss = -245.6 (0.67)\n",
            "epoch: 16388, dloss = 3.2, gloss = -2508.6, total_loss = -249.2 (0.68)\n",
            "epoch: 16389, dloss = 3.0, gloss = -2459.5, total_loss = -244.5 (0.66)\n",
            "epoch: 16390, dloss = 3.3, gloss = -2436.5, total_loss = -242.0 (0.69)\n",
            "epoch: 16391, dloss = 3.4, gloss = -2475.2, total_loss = -245.8 (0.69)\n",
            "epoch: 16392, dloss = 0.6, gloss = -2472.0, total_loss = -246.9 (0.71)\n",
            "epoch: 16393, dloss = 6.0, gloss = -2453.1, total_loss = -242.3 (0.87)\n",
            "epoch: 16394, dloss = 3.5, gloss = -2423.9, total_loss = -240.6 (0.80)\n",
            "epoch: 16395, dloss = 4.0, gloss = -2411.5, total_loss = -239.2 (0.68)\n",
            "epoch: 16396, dloss = 3.6, gloss = -2407.2, total_loss = -238.9 (0.69)\n",
            "epoch: 16397, dloss = 2.7, gloss = -2368.0, total_loss = -235.4 (0.67)\n",
            "epoch: 16398, dloss = -2.6, gloss = -2371.9, total_loss = -238.5 (0.68)\n",
            "epoch: 16399, dloss = 1.3, gloss = -2437.2, total_loss = -243.1 (0.82)\n",
            "epoch: 16400, dloss = -2.0, gloss = -2364.1, total_loss = -237.4 (1.01)\n",
            "epoch: 16401, dloss = 3.5, gloss = -2401.7, total_loss = -238.4 (0.67)\n",
            "epoch: 16402, dloss = 1.5, gloss = -2403.2, total_loss = -239.6 (0.71)\n",
            "epoch: 16403, dloss = 1.5, gloss = -2387.3, total_loss = -238.0 (0.67)\n",
            "epoch: 16404, dloss = 1.2, gloss = -2406.0, total_loss = -240.0 (0.69)\n",
            "epoch: 16405, dloss = 3.8, gloss = -2424.5, total_loss = -240.6 (0.70)\n",
            "epoch: 16406, dloss = 1.6, gloss = -2417.8, total_loss = -241.0 (1.07)\n",
            "epoch: 16407, dloss = 1.5, gloss = -2434.6, total_loss = -242.7 (1.26)\n",
            "epoch: 16408, dloss = -0.8, gloss = -2417.8, total_loss = -242.2 (1.47)\n",
            "epoch: 16409, dloss = 1.7, gloss = -2464.6, total_loss = -245.6 (1.12)\n",
            "epoch: 16410, dloss = 4.7, gloss = -2440.3, total_loss = -241.7 (0.96)\n",
            "epoch: 16411, dloss = 3.5, gloss = -2426.4, total_loss = -240.9 (0.67)\n",
            "epoch: 16412, dloss = 2.1, gloss = -2414.8, total_loss = -240.4 (0.70)\n",
            "epoch: 16413, dloss = 1.4, gloss = -2350.4, total_loss = -234.3 (0.69)\n",
            "epoch: 16414, dloss = 1.9, gloss = -2324.7, total_loss = -231.5 (0.67)\n",
            "epoch: 16415, dloss = 3.1, gloss = -2311.5, total_loss = -229.6 (0.68)\n",
            "epoch: 16416, dloss = 2.8, gloss = -2378.8, total_loss = -236.5 (0.69)\n",
            "epoch: 16417, dloss = 2.2, gloss = -2344.7, total_loss = -233.4 (0.69)\n",
            "epoch: 16418, dloss = 3.7, gloss = -2371.6, total_loss = -235.3 (0.67)\n",
            "epoch: 16419, dloss = 0.5, gloss = -2364.4, total_loss = -236.2 (0.68)\n",
            "epoch: 16420, dloss = 2.4, gloss = -2443.7, total_loss = -243.2 (0.71)\n",
            "epoch: 16421, dloss = 5.0, gloss = -2372.8, total_loss = -234.8 (0.87)\n",
            "epoch: 16422, dloss = 2.6, gloss = -2328.2, total_loss = -231.5 (0.84)\n",
            "epoch: 16423, dloss = 1.6, gloss = -2417.2, total_loss = -240.9 (0.67)\n",
            "epoch: 16424, dloss = 2.8, gloss = -2331.9, total_loss = -231.8 (0.67)\n",
            "epoch: 16425, dloss = 3.1, gloss = -2329.1, total_loss = -231.3 (0.66)\n",
            "epoch: 16426, dloss = -0.1, gloss = -2320.1, total_loss = -232.1 (0.67)\n",
            "epoch: 16427, dloss = 1.4, gloss = -2264.8, total_loss = -225.8 (0.70)\n",
            "epoch: 16428, dloss = 0.2, gloss = -2232.1, total_loss = -223.1 (0.67)\n",
            "epoch: 16429, dloss = 1.7, gloss = -2293.2, total_loss = -228.5 (0.67)\n",
            "epoch: 16430, dloss = 2.8, gloss = -2288.7, total_loss = -227.5 (0.70)\n",
            "epoch: 16431, dloss = 2.9, gloss = -2364.9, total_loss = -235.0 (0.66)\n",
            "epoch: 16432, dloss = 5.3, gloss = -2351.6, total_loss = -232.5 (0.65)\n",
            "epoch: 16433, dloss = 0.6, gloss = -2326.3, total_loss = -232.3 (0.68)\n",
            "epoch: 16434, dloss = 4.6, gloss = -2320.6, total_loss = -229.8 (0.68)\n",
            "epoch: 16435, dloss = 1.1, gloss = -2325.2, total_loss = -231.9 (0.68)\n",
            "epoch: 16436, dloss = 3.3, gloss = -2367.9, total_loss = -235.2 (0.68)\n",
            "epoch: 16437, dloss = 2.5, gloss = -2328.9, total_loss = -231.6 (0.96)\n",
            "epoch: 16438, dloss = 3.9, gloss = -2383.9, total_loss = -236.4 (0.77)\n",
            "epoch: 16439, dloss = 0.1, gloss = -2365.2, total_loss = -236.5 (0.70)\n",
            "epoch: 16440, dloss = -0.3, gloss = -2367.9, total_loss = -236.9 (0.69)\n",
            "epoch: 16441, dloss = 2.3, gloss = -2349.8, total_loss = -233.8 (0.70)\n",
            "epoch: 16442, dloss = 1.2, gloss = -2322.8, total_loss = -231.7 (0.69)\n",
            "epoch: 16443, dloss = 3.8, gloss = -2324.4, total_loss = -230.5 (0.66)\n",
            "epoch: 16444, dloss = 4.8, gloss = -2416.3, total_loss = -239.2 (0.68)\n",
            "epoch: 16445, dloss = 2.7, gloss = -2334.3, total_loss = -232.1 (0.68)\n",
            "epoch: 16446, dloss = 3.0, gloss = -2354.2, total_loss = -233.9 (0.68)\n",
            "epoch: 16447, dloss = -0.1, gloss = -2324.2, total_loss = -232.5 (0.69)\n",
            "epoch: 16448, dloss = 3.0, gloss = -2393.5, total_loss = -237.9 (0.68)\n",
            "epoch: 16449, dloss = -0.1, gloss = -2399.7, total_loss = -240.0 (0.67)\n",
            "epoch: 16450, dloss = -0.9, gloss = -2425.1, total_loss = -243.0 (0.67)\n",
            "epoch: 16451, dloss = -0.5, gloss = -2418.8, total_loss = -242.1 (0.68)\n",
            "epoch: 16452, dloss = 4.2, gloss = -2403.1, total_loss = -238.2 (0.74)\n",
            "epoch: 16453, dloss = 2.5, gloss = -2409.7, total_loss = -239.7 (0.97)\n",
            "epoch: 16454, dloss = -9.5, gloss = -2413.4, total_loss = -246.1 (0.68)\n",
            "epoch: 16455, dloss = 0.7, gloss = -2435.5, total_loss = -243.2 (0.68)\n",
            "epoch: 16456, dloss = 2.6, gloss = -2437.8, total_loss = -242.5 (0.68)\n",
            "epoch: 16457, dloss = 0.6, gloss = -2385.8, total_loss = -238.3 (0.68)\n",
            "epoch: 16458, dloss = 1.8, gloss = -2361.3, total_loss = -235.2 (0.69)\n",
            "epoch: 16459, dloss = 3.5, gloss = -2365.3, total_loss = -234.8 (0.67)\n",
            "epoch: 16460, dloss = -0.9, gloss = -2347.5, total_loss = -235.2 (0.69)\n",
            "epoch: 16461, dloss = 3.8, gloss = -2408.8, total_loss = -239.0 (0.72)\n",
            "epoch: 16462, dloss = 3.2, gloss = -2329.9, total_loss = -231.4 (0.70)\n",
            "epoch: 16463, dloss = -0.5, gloss = -2365.9, total_loss = -236.8 (0.68)\n",
            "epoch: 16464, dloss = 4.8, gloss = -2393.0, total_loss = -236.9 (0.68)\n",
            "epoch: 16465, dloss = 3.0, gloss = -2429.5, total_loss = -241.4 (0.68)\n",
            "epoch: 16466, dloss = 4.1, gloss = -2404.1, total_loss = -238.4 (0.69)\n",
            "epoch: 16467, dloss = 1.3, gloss = -2423.9, total_loss = -241.7 (0.68)\n",
            "epoch: 16468, dloss = 4.3, gloss = -2457.4, total_loss = -243.6 (0.90)\n",
            "epoch: 16469, dloss = 2.9, gloss = -2395.6, total_loss = -238.1 (0.83)\n",
            "epoch: 16470, dloss = 1.6, gloss = -2371.1, total_loss = -236.3 (0.67)\n",
            "epoch: 16471, dloss = 0.5, gloss = -2357.6, total_loss = -235.5 (0.71)\n",
            "epoch: 16472, dloss = 2.5, gloss = -2387.3, total_loss = -237.5 (0.68)\n",
            "epoch: 16473, dloss = 3.8, gloss = -2291.6, total_loss = -227.3 (0.69)\n",
            "epoch: 16474, dloss = 1.7, gloss = -2316.2, total_loss = -230.8 (0.69)\n",
            "epoch: 16475, dloss = 0.9, gloss = -2379.0, total_loss = -237.4 (0.69)\n",
            "epoch: 16476, dloss = 3.0, gloss = -2251.1, total_loss = -223.6 (0.68)\n",
            "epoch: 16477, dloss = 0.0, gloss = -2405.2, total_loss = -240.5 (0.69)\n",
            "epoch: 16478, dloss = -0.8, gloss = -2338.5, total_loss = -234.3 (0.71)\n",
            "epoch: 16479, dloss = 2.2, gloss = -2345.2, total_loss = -233.4 (0.68)\n",
            "epoch: 16480, dloss = -1.2, gloss = -2275.4, total_loss = -228.1 (0.69)\n",
            "epoch: 16481, dloss = 5.8, gloss = -2384.4, total_loss = -235.5 (0.67)\n",
            "epoch: 16482, dloss = 0.5, gloss = -2414.6, total_loss = -241.2 (0.70)\n",
            "epoch: 16483, dloss = 3.0, gloss = -2395.2, total_loss = -238.0 (0.73)\n",
            "epoch: 16484, dloss = 3.6, gloss = -2414.0, total_loss = -239.6 (1.02)\n",
            "epoch: 16485, dloss = 6.2, gloss = -2380.7, total_loss = -235.0 (0.69)\n",
            "epoch: 16486, dloss = 1.4, gloss = -2375.6, total_loss = -236.9 (0.70)\n",
            "epoch: 16487, dloss = 0.7, gloss = -2356.7, total_loss = -235.3 (0.69)\n",
            "epoch: 16488, dloss = 0.3, gloss = -2411.9, total_loss = -241.0 (0.68)\n",
            "epoch: 16489, dloss = 1.5, gloss = -2409.8, total_loss = -240.2 (0.69)\n",
            "epoch: 16490, dloss = 8.1, gloss = -2364.8, total_loss = -232.4 (0.69)\n",
            "epoch: 16491, dloss = 3.9, gloss = -2368.5, total_loss = -234.9 (0.68)\n",
            "epoch: 16492, dloss = 0.8, gloss = -2386.6, total_loss = -238.3 (0.70)\n",
            "epoch: 16493, dloss = 3.3, gloss = -2401.4, total_loss = -238.5 (0.67)\n",
            "epoch: 16494, dloss = 1.4, gloss = -2376.5, total_loss = -236.9 (0.70)\n",
            "epoch: 16495, dloss = -3.7, gloss = -2295.9, total_loss = -231.4 (0.70)\n",
            "epoch: 16496, dloss = 0.6, gloss = -2364.4, total_loss = -236.2 (0.68)\n",
            "epoch: 16497, dloss = 1.4, gloss = -2311.9, total_loss = -230.5 (0.67)\n",
            "epoch: 16498, dloss = -2.3, gloss = -2402.6, total_loss = -241.4 (0.69)\n",
            "epoch: 16499, dloss = 3.4, gloss = -2410.1, total_loss = -239.3 (0.88)\n",
            "epoch: 16500, dloss = 2.8, gloss = -2399.2, total_loss = -238.5 (0.82)\n",
            "epoch: 16501, dloss = 3.6, gloss = -2370.1, total_loss = -235.2 (0.69)\n",
            "epoch: 16502, dloss = -0.6, gloss = -2400.4, total_loss = -240.3 (0.69)\n",
            "epoch: 16503, dloss = 3.3, gloss = -2379.2, total_loss = -236.3 (0.68)\n",
            "epoch: 16504, dloss = 1.9, gloss = -2415.2, total_loss = -240.6 (0.68)\n",
            "epoch: 16505, dloss = -2.6, gloss = -2332.2, total_loss = -234.5 (0.68)\n",
            "epoch: 16506, dloss = -0.5, gloss = -2384.2, total_loss = -238.6 (0.68)\n",
            "epoch: 16507, dloss = 2.0, gloss = -2378.2, total_loss = -236.8 (0.69)\n",
            "epoch: 16508, dloss = 2.6, gloss = -2412.4, total_loss = -239.9 (0.70)\n",
            "epoch: 16509, dloss = -2.5, gloss = -2343.1, total_loss = -235.6 (0.69)\n",
            "epoch: 16510, dloss = 1.7, gloss = -2363.9, total_loss = -235.6 (0.69)\n",
            "epoch: 16511, dloss = 5.7, gloss = -2408.3, total_loss = -238.0 (0.69)\n",
            "epoch: 16512, dloss = 20.0, gloss = -2445.1, total_loss = -234.5 (0.69)\n",
            "epoch: 16513, dloss = 18.3, gloss = -2563.8, total_loss = -247.2 (0.69)\n",
            "epoch: 16514, dloss = 16.9, gloss = -2633.6, total_loss = -254.9 (0.71)\n",
            "epoch: 16515, dloss = 24.0, gloss = -2629.5, total_loss = -251.0 (1.03)\n",
            "epoch: 16516, dloss = 16.9, gloss = -2642.3, total_loss = -255.8 (0.67)\n",
            "epoch: 16517, dloss = 12.3, gloss = -2717.7, total_loss = -265.6 (0.69)\n",
            "epoch: 16518, dloss = 11.0, gloss = -2733.1, total_loss = -267.8 (0.70)\n",
            "epoch: 16519, dloss = 8.2, gloss = -2735.8, total_loss = -269.5 (0.67)\n",
            "epoch: 16520, dloss = 10.7, gloss = -2779.3, total_loss = -272.6 (0.70)\n",
            "epoch: 16521, dloss = 8.4, gloss = -2832.4, total_loss = -279.0 (0.69)\n",
            "epoch: 16522, dloss = 4.4, gloss = -2800.1, total_loss = -277.8 (0.69)\n",
            "epoch: 16523, dloss = 11.4, gloss = -2787.1, total_loss = -273.0 (0.68)\n",
            "epoch: 16524, dloss = 9.7, gloss = -2807.6, total_loss = -275.9 (0.69)\n",
            "epoch: 16525, dloss = 5.7, gloss = -2811.2, total_loss = -278.3 (0.71)\n",
            "epoch: 16526, dloss = 6.1, gloss = -2819.9, total_loss = -279.0 (0.68)\n",
            "epoch: 16527, dloss = 3.7, gloss = -2744.0, total_loss = -272.6 (0.69)\n",
            "epoch: 16528, dloss = 5.1, gloss = -2730.9, total_loss = -270.5 (0.69)\n",
            "epoch: 16529, dloss = 4.4, gloss = -2722.7, total_loss = -270.1 (0.69)\n",
            "epoch: 16530, dloss = 7.8, gloss = -2758.9, total_loss = -272.0 (0.86)\n",
            "epoch: 16531, dloss = 6.2, gloss = -2712.4, total_loss = -268.2 (0.89)\n",
            "epoch: 16532, dloss = 6.7, gloss = -2717.3, total_loss = -268.4 (0.67)\n",
            "epoch: 16533, dloss = 7.1, gloss = -2819.4, total_loss = -278.4 (0.66)\n",
            "epoch: 16534, dloss = 6.1, gloss = -2842.7, total_loss = -281.2 (0.70)\n",
            "epoch: 16535, dloss = 5.7, gloss = -2839.6, total_loss = -281.1 (0.69)\n",
            "epoch: 16536, dloss = 5.1, gloss = -2778.0, total_loss = -275.2 (0.67)\n",
            "epoch: 16537, dloss = 2.0, gloss = -2728.6, total_loss = -271.9 (0.68)\n",
            "epoch: 16538, dloss = 5.4, gloss = -2786.5, total_loss = -276.0 (0.71)\n",
            "epoch: 16539, dloss = 6.2, gloss = -2756.7, total_loss = -272.6 (0.68)\n",
            "epoch: 16540, dloss = 4.8, gloss = -2642.7, total_loss = -261.9 (0.68)\n",
            "epoch: 16541, dloss = 4.6, gloss = -2671.7, total_loss = -264.9 (0.67)\n",
            "epoch: 16542, dloss = 2.1, gloss = -2615.3, total_loss = -260.5 (0.68)\n",
            "epoch: 16543, dloss = 5.4, gloss = -2683.8, total_loss = -265.7 (0.68)\n",
            "epoch: 16544, dloss = 4.8, gloss = -2655.4, total_loss = -263.1 (0.67)\n",
            "epoch: 16545, dloss = 7.4, gloss = -2703.7, total_loss = -266.7 (0.68)\n",
            "epoch: 16546, dloss = 2.4, gloss = -2694.4, total_loss = -268.3 (0.95)\n",
            "epoch: 16547, dloss = 6.1, gloss = -2701.9, total_loss = -267.1 (0.78)\n",
            "epoch: 16548, dloss = 6.6, gloss = -2701.4, total_loss = -266.8 (0.68)\n",
            "epoch: 16549, dloss = 6.7, gloss = -2701.8, total_loss = -266.8 (0.68)\n",
            "epoch: 16550, dloss = 5.5, gloss = -2682.1, total_loss = -265.5 (0.65)\n",
            "epoch: 16551, dloss = 3.6, gloss = -2667.5, total_loss = -264.9 (0.68)\n",
            "epoch: 16552, dloss = 4.3, gloss = -2723.7, total_loss = -270.2 (0.66)\n",
            "epoch: 16553, dloss = 5.7, gloss = -2717.1, total_loss = -268.9 (0.68)\n",
            "epoch: 16554, dloss = 3.7, gloss = -2757.6, total_loss = -273.9 (0.69)\n",
            "epoch: 16555, dloss = 2.6, gloss = -2670.1, total_loss = -265.7 (0.66)\n",
            "epoch: 16556, dloss = 5.7, gloss = -2734.2, total_loss = -270.6 (0.66)\n",
            "epoch: 16557, dloss = 5.0, gloss = -2739.9, total_loss = -271.5 (0.69)\n",
            "epoch: 16558, dloss = 3.6, gloss = -2746.9, total_loss = -272.9 (0.69)\n",
            "epoch: 16559, dloss = 3.2, gloss = -2717.4, total_loss = -270.2 (0.70)\n",
            "epoch: 16560, dloss = 4.1, gloss = -2698.5, total_loss = -267.8 (1.07)\n",
            "epoch: 16561, dloss = 5.1, gloss = -2722.5, total_loss = -269.7 (0.90)\n",
            "epoch: 16562, dloss = 2.9, gloss = -2700.2, total_loss = -268.6 (0.84)\n",
            "epoch: 16563, dloss = 4.8, gloss = -2673.7, total_loss = -265.0 (0.67)\n",
            "epoch: 16564, dloss = 2.5, gloss = -2720.5, total_loss = -270.8 (0.67)\n",
            "epoch: 16565, dloss = -0.1, gloss = -2674.6, total_loss = -267.5 (0.66)\n",
            "epoch: 16566, dloss = 4.0, gloss = -2680.8, total_loss = -266.1 (0.68)\n",
            "epoch: 16567, dloss = 2.1, gloss = -2698.3, total_loss = -268.8 (0.71)\n",
            "epoch: 16568, dloss = 4.4, gloss = -2725.7, total_loss = -270.4 (0.69)\n",
            "epoch: 16569, dloss = 4.0, gloss = -2701.5, total_loss = -268.1 (0.69)\n",
            "epoch: 16570, dloss = 4.8, gloss = -2746.3, total_loss = -272.3 (0.69)\n",
            "epoch: 16571, dloss = 5.8, gloss = -2695.8, total_loss = -266.7 (0.68)\n",
            "epoch: 16572, dloss = 4.8, gloss = -2692.5, total_loss = -266.9 (0.88)\n",
            "epoch: 16573, dloss = 4.8, gloss = -2690.1, total_loss = -266.6 (0.84)\n",
            "epoch: 16574, dloss = 2.1, gloss = -2660.9, total_loss = -265.1 (0.68)\n",
            "epoch: 16575, dloss = 3.3, gloss = -2677.1, total_loss = -266.0 (0.68)\n",
            "epoch: 16576, dloss = 4.8, gloss = -2656.9, total_loss = -263.3 (0.87)\n",
            "epoch: 16577, dloss = 4.4, gloss = -2691.6, total_loss = -267.0 (0.82)\n",
            "epoch: 16578, dloss = 5.9, gloss = -2681.7, total_loss = -265.2 (0.69)\n",
            "epoch: 16579, dloss = 4.8, gloss = -2688.5, total_loss = -266.4 (0.70)\n",
            "epoch: 16580, dloss = 5.1, gloss = -2704.5, total_loss = -267.9 (0.68)\n",
            "epoch: 16581, dloss = 4.4, gloss = -2661.2, total_loss = -263.9 (0.69)\n",
            "epoch: 16582, dloss = 5.2, gloss = -2663.9, total_loss = -263.8 (0.68)\n",
            "epoch: 16583, dloss = 3.7, gloss = -2656.2, total_loss = -263.8 (0.67)\n",
            "epoch: 16584, dloss = 3.9, gloss = -2672.3, total_loss = -265.3 (0.67)\n",
            "epoch: 16585, dloss = 3.8, gloss = -2672.9, total_loss = -265.4 (0.69)\n",
            "epoch: 16586, dloss = 3.9, gloss = -2689.9, total_loss = -267.0 (0.67)\n",
            "epoch: 16587, dloss = 2.4, gloss = -2670.2, total_loss = -265.8 (0.66)\n",
            "epoch: 16588, dloss = 2.3, gloss = -2702.8, total_loss = -269.1 (0.66)\n",
            "epoch: 16589, dloss = 4.7, gloss = -2724.1, total_loss = -270.1 (0.69)\n",
            "epoch: 16590, dloss = 2.9, gloss = -2686.4, total_loss = -267.2 (0.70)\n",
            "epoch: 16591, dloss = -0.5, gloss = -2657.8, total_loss = -266.0 (0.68)\n",
            "epoch: 16592, dloss = 0.7, gloss = -2667.1, total_loss = -266.4 (0.99)\n",
            "epoch: 16593, dloss = 3.3, gloss = -2659.7, total_loss = -264.3 (0.75)\n",
            "epoch: 16594, dloss = 2.2, gloss = -2658.6, total_loss = -264.8 (0.68)\n",
            "epoch: 16595, dloss = 3.0, gloss = -2652.8, total_loss = -263.8 (0.69)\n",
            "epoch: 16596, dloss = 1.9, gloss = -2599.3, total_loss = -259.0 (0.66)\n",
            "epoch: 16597, dloss = 3.9, gloss = -2563.2, total_loss = -254.4 (0.69)\n",
            "epoch: 16598, dloss = 1.7, gloss = -2570.0, total_loss = -256.2 (0.68)\n",
            "epoch: 16599, dloss = -0.2, gloss = -2540.5, total_loss = -254.1 (0.70)\n",
            "epoch: 16600, dloss = 3.9, gloss = -2573.4, total_loss = -255.4 (0.69)\n",
            "epoch: 16601, dloss = 4.8, gloss = -2582.9, total_loss = -255.9 (0.69)\n",
            "epoch: 16602, dloss = 3.8, gloss = -2566.0, total_loss = -254.7 (0.70)\n",
            "epoch: 16603, dloss = 3.1, gloss = -2565.6, total_loss = -255.0 (0.70)\n",
            "epoch: 16604, dloss = 1.5, gloss = -2485.3, total_loss = -247.8 (0.69)\n",
            "epoch: 16605, dloss = 4.3, gloss = -2578.8, total_loss = -255.7 (0.68)\n",
            "epoch: 16606, dloss = 3.1, gloss = -2635.9, total_loss = -262.1 (0.67)\n",
            "epoch: 16607, dloss = 4.0, gloss = -2620.1, total_loss = -260.0 (0.77)\n",
            "epoch: 16608, dloss = 0.4, gloss = -2562.6, total_loss = -256.0 (0.93)\n",
            "epoch: 16609, dloss = 1.7, gloss = -2540.6, total_loss = -253.2 (0.69)\n",
            "epoch: 16610, dloss = 3.0, gloss = -2617.1, total_loss = -260.2 (0.68)\n",
            "epoch: 16611, dloss = 3.8, gloss = -2620.2, total_loss = -260.1 (0.68)\n",
            "epoch: 16612, dloss = 4.4, gloss = -2631.8, total_loss = -261.0 (0.69)\n",
            "epoch: 16613, dloss = -1.9, gloss = -2643.6, total_loss = -265.3 (0.69)\n",
            "epoch: 16614, dloss = 5.8, gloss = -2578.3, total_loss = -254.9 (0.69)\n",
            "epoch: 16615, dloss = 3.6, gloss = -2553.3, total_loss = -253.5 (0.68)\n",
            "epoch: 16616, dloss = 3.7, gloss = -2574.6, total_loss = -255.6 (0.68)\n",
            "epoch: 16617, dloss = 2.4, gloss = -2555.5, total_loss = -254.4 (0.70)\n",
            "epoch: 16618, dloss = 1.7, gloss = -2550.5, total_loss = -254.2 (0.72)\n",
            "epoch: 16619, dloss = 2.9, gloss = -2582.7, total_loss = -256.8 (0.68)\n",
            "epoch: 16620, dloss = 6.1, gloss = -2594.0, total_loss = -256.3 (0.70)\n",
            "epoch: 16621, dloss = 4.8, gloss = -2624.5, total_loss = -260.0 (0.70)\n",
            "epoch: 16622, dloss = 2.9, gloss = -2608.0, total_loss = -259.3 (0.68)\n",
            "epoch: 16623, dloss = 3.0, gloss = -2574.3, total_loss = -255.9 (0.95)\n",
            "epoch: 16624, dloss = 0.4, gloss = -2611.4, total_loss = -260.9 (0.79)\n",
            "epoch: 16625, dloss = -0.4, gloss = -2598.7, total_loss = -260.1 (0.69)\n",
            "epoch: 16626, dloss = 1.6, gloss = -2617.2, total_loss = -260.9 (0.67)\n",
            "epoch: 16627, dloss = 2.7, gloss = -2597.5, total_loss = -258.4 (0.67)\n",
            "epoch: 16628, dloss = 5.3, gloss = -2464.7, total_loss = -243.8 (0.67)\n",
            "epoch: 16629, dloss = 2.6, gloss = -2525.8, total_loss = -251.3 (0.70)\n",
            "epoch: 16630, dloss = 0.6, gloss = -2498.8, total_loss = -249.6 (0.70)\n",
            "epoch: 16631, dloss = 2.8, gloss = -2489.7, total_loss = -247.6 (0.70)\n",
            "epoch: 16632, dloss = 1.9, gloss = -2505.1, total_loss = -249.6 (0.67)\n",
            "epoch: 16633, dloss = 1.3, gloss = -2447.3, total_loss = -244.1 (0.66)\n",
            "epoch: 16634, dloss = -0.5, gloss = -2478.8, total_loss = -248.1 (0.68)\n",
            "epoch: 16635, dloss = 3.0, gloss = -2444.0, total_loss = -242.9 (0.68)\n",
            "epoch: 16636, dloss = 2.2, gloss = -2503.7, total_loss = -249.3 (0.65)\n",
            "epoch: 16637, dloss = 4.7, gloss = -2501.2, total_loss = -247.8 (0.69)\n",
            "epoch: 16638, dloss = 2.5, gloss = -2502.2, total_loss = -249.0 (0.71)\n",
            "epoch: 16639, dloss = 1.6, gloss = -2510.8, total_loss = -250.3 (1.00)\n",
            "epoch: 16640, dloss = 3.9, gloss = -2555.2, total_loss = -253.6 (0.68)\n",
            "epoch: 16641, dloss = 0.1, gloss = -2490.2, total_loss = -249.0 (0.67)\n",
            "epoch: 16642, dloss = 4.0, gloss = -2478.0, total_loss = -245.8 (0.68)\n",
            "epoch: 16643, dloss = 2.8, gloss = -2455.5, total_loss = -244.1 (0.70)\n",
            "epoch: 16644, dloss = 2.7, gloss = -2405.9, total_loss = -239.2 (0.67)\n",
            "epoch: 16645, dloss = 2.7, gloss = -2448.8, total_loss = -243.5 (0.69)\n",
            "epoch: 16646, dloss = 3.5, gloss = -2434.0, total_loss = -241.7 (0.69)\n",
            "epoch: 16647, dloss = 3.5, gloss = -2426.7, total_loss = -240.9 (0.70)\n",
            "epoch: 16648, dloss = 3.4, gloss = -2412.5, total_loss = -239.5 (0.68)\n",
            "epoch: 16649, dloss = 5.8, gloss = -2483.3, total_loss = -245.4 (0.70)\n",
            "epoch: 16650, dloss = 1.9, gloss = -2463.6, total_loss = -245.4 (0.69)\n",
            "epoch: 16651, dloss = 2.5, gloss = -2488.1, total_loss = -247.6 (0.68)\n",
            "epoch: 16652, dloss = 4.5, gloss = -2493.3, total_loss = -247.1 (0.69)\n",
            "epoch: 16653, dloss = 1.8, gloss = -2514.5, total_loss = -250.5 (0.69)\n",
            "epoch: 16654, dloss = 2.8, gloss = -2480.5, total_loss = -246.6 (0.92)\n",
            "epoch: 16655, dloss = 3.6, gloss = -2480.0, total_loss = -246.2 (0.86)\n",
            "epoch: 16656, dloss = 2.6, gloss = -2494.2, total_loss = -248.1 (0.69)\n",
            "epoch: 16657, dloss = 0.7, gloss = -2501.7, total_loss = -249.8 (0.69)\n",
            "epoch: 16658, dloss = -1.4, gloss = -2495.8, total_loss = -250.3 (0.70)\n",
            "epoch: 16659, dloss = -0.7, gloss = -2470.7, total_loss = -247.4 (0.72)\n",
            "epoch: 16660, dloss = 4.2, gloss = -2507.0, total_loss = -248.6 (0.68)\n",
            "epoch: 16661, dloss = 3.3, gloss = -2451.1, total_loss = -243.5 (0.68)\n",
            "epoch: 16662, dloss = 4.3, gloss = -2470.6, total_loss = -244.9 (0.67)\n",
            "epoch: 16663, dloss = 4.4, gloss = -2429.7, total_loss = -240.8 (0.70)\n",
            "epoch: 16664, dloss = 3.7, gloss = -2410.7, total_loss = -239.2 (0.68)\n",
            "epoch: 16665, dloss = 3.8, gloss = -2452.8, total_loss = -243.4 (0.66)\n",
            "epoch: 16666, dloss = 2.7, gloss = -2405.5, total_loss = -239.2 (0.70)\n",
            "epoch: 16667, dloss = 1.7, gloss = -2414.0, total_loss = -240.5 (0.68)\n",
            "epoch: 16668, dloss = -0.6, gloss = -2379.2, total_loss = -238.2 (0.65)\n",
            "epoch: 16669, dloss = 0.2, gloss = -2393.4, total_loss = -239.2 (0.70)\n",
            "epoch: 16670, dloss = 5.4, gloss = -2394.2, total_loss = -236.7 (1.02)\n",
            "epoch: 16671, dloss = 3.3, gloss = -2422.7, total_loss = -240.6 (0.69)\n",
            "epoch: 16672, dloss = 2.9, gloss = -2402.0, total_loss = -238.7 (0.68)\n",
            "epoch: 16673, dloss = 1.0, gloss = -2419.1, total_loss = -241.4 (0.70)\n",
            "epoch: 16674, dloss = 1.2, gloss = -2428.5, total_loss = -242.3 (0.68)\n",
            "epoch: 16675, dloss = 0.7, gloss = -2475.9, total_loss = -247.2 (0.67)\n",
            "epoch: 16676, dloss = 0.9, gloss = -2472.4, total_loss = -246.8 (0.68)\n",
            "epoch: 16677, dloss = 4.1, gloss = -2501.8, total_loss = -248.1 (0.69)\n",
            "epoch: 16678, dloss = 0.5, gloss = -2462.4, total_loss = -246.0 (0.69)\n",
            "epoch: 16679, dloss = 3.4, gloss = -2459.8, total_loss = -244.3 (0.67)\n",
            "epoch: 16680, dloss = 2.8, gloss = -2456.5, total_loss = -244.2 (0.69)\n",
            "epoch: 16681, dloss = 1.6, gloss = -2461.2, total_loss = -245.3 (0.67)\n",
            "epoch: 16682, dloss = -0.1, gloss = -2438.0, total_loss = -243.9 (0.69)\n",
            "epoch: 16683, dloss = 0.5, gloss = -2412.2, total_loss = -241.0 (0.69)\n",
            "epoch: 16684, dloss = -1.1, gloss = -2423.5, total_loss = -242.9 (0.68)\n",
            "epoch: 16685, dloss = -0.4, gloss = -2417.8, total_loss = -242.0 (0.84)\n",
            "epoch: 16686, dloss = 4.2, gloss = -2414.7, total_loss = -239.4 (1.06)\n",
            "epoch: 16687, dloss = 2.8, gloss = -2427.1, total_loss = -241.3 (0.97)\n",
            "epoch: 16688, dloss = 2.2, gloss = -2398.9, total_loss = -238.8 (0.70)\n",
            "epoch: 16689, dloss = 2.0, gloss = -2345.6, total_loss = -233.6 (0.68)\n",
            "epoch: 16690, dloss = 1.0, gloss = -2331.0, total_loss = -232.6 (0.70)\n",
            "epoch: 16691, dloss = 1.0, gloss = -2336.2, total_loss = -233.1 (0.69)\n",
            "epoch: 16692, dloss = 1.6, gloss = -2391.1, total_loss = -238.3 (0.69)\n",
            "epoch: 16693, dloss = 3.3, gloss = -2374.9, total_loss = -235.9 (0.67)\n",
            "epoch: 16694, dloss = 1.3, gloss = -2360.0, total_loss = -235.3 (0.70)\n",
            "epoch: 16695, dloss = 2.3, gloss = -2308.7, total_loss = -229.7 (0.69)\n",
            "epoch: 16696, dloss = 1.5, gloss = -2292.7, total_loss = -228.5 (0.68)\n",
            "epoch: 16697, dloss = 0.8, gloss = -2335.2, total_loss = -233.1 (0.70)\n",
            "epoch: 16698, dloss = 2.2, gloss = -2241.0, total_loss = -223.0 (0.69)\n",
            "epoch: 16699, dloss = 4.2, gloss = -2261.1, total_loss = -224.0 (0.67)\n",
            "epoch: 16700, dloss = 0.2, gloss = -2279.0, total_loss = -227.8 (0.84)\n",
            "epoch: 16701, dloss = 2.8, gloss = -2226.1, total_loss = -221.2 (0.87)\n",
            "epoch: 16702, dloss = -1.2, gloss = -2182.5, total_loss = -218.8 (0.70)\n",
            "epoch: 16703, dloss = 1.1, gloss = -2259.1, total_loss = -225.4 (0.68)\n",
            "epoch: 16704, dloss = 2.8, gloss = -2217.4, total_loss = -220.3 (0.69)\n",
            "epoch: 16705, dloss = 1.6, gloss = -2280.2, total_loss = -227.2 (0.75)\n",
            "epoch: 16706, dloss = 1.3, gloss = -2391.0, total_loss = -238.4 (0.69)\n",
            "epoch: 16707, dloss = 2.6, gloss = -2285.7, total_loss = -227.3 (0.68)\n",
            "epoch: 16708, dloss = 0.8, gloss = -2306.3, total_loss = -230.2 (0.69)\n",
            "epoch: 16709, dloss = -0.3, gloss = -2279.8, total_loss = -228.1 (0.68)\n",
            "epoch: 16710, dloss = 0.0, gloss = -2282.7, total_loss = -228.3 (0.67)\n",
            "epoch: 16711, dloss = -5.3, gloss = -2284.7, total_loss = -231.1 (0.69)\n",
            "epoch: 16712, dloss = 0.8, gloss = -2314.4, total_loss = -231.1 (0.68)\n",
            "epoch: 16713, dloss = 1.8, gloss = -2371.3, total_loss = -236.3 (0.68)\n",
            "epoch: 16714, dloss = 2.6, gloss = -2359.1, total_loss = -234.6 (0.69)\n",
            "epoch: 16715, dloss = -1.6, gloss = -2322.0, total_loss = -233.0 (0.67)\n",
            "epoch: 16716, dloss = 0.6, gloss = -2268.5, total_loss = -226.6 (1.01)\n",
            "epoch: 16717, dloss = 3.7, gloss = -2271.3, total_loss = -225.3 (0.73)\n",
            "epoch: 16718, dloss = 1.5, gloss = -2202.6, total_loss = -219.5 (0.69)\n",
            "epoch: 16719, dloss = 2.9, gloss = -2279.3, total_loss = -226.5 (0.70)\n",
            "epoch: 16720, dloss = 0.0, gloss = -2343.2, total_loss = -234.3 (0.70)\n",
            "epoch: 16721, dloss = 3.2, gloss = -2182.4, total_loss = -216.6 (0.67)\n",
            "epoch: 16722, dloss = -2.8, gloss = -2224.2, total_loss = -223.8 (0.69)\n",
            "epoch: 16723, dloss = 3.3, gloss = -2200.6, total_loss = -218.4 (0.68)\n",
            "epoch: 16724, dloss = -1.7, gloss = -2326.5, total_loss = -233.5 (0.69)\n",
            "epoch: 16725, dloss = 1.3, gloss = -2291.3, total_loss = -228.5 (0.70)\n",
            "epoch: 16726, dloss = 1.5, gloss = -2365.5, total_loss = -235.8 (0.68)\n",
            "epoch: 16727, dloss = 1.0, gloss = -2287.3, total_loss = -228.2 (0.67)\n",
            "epoch: 16728, dloss = 3.0, gloss = -2269.8, total_loss = -225.5 (0.69)\n",
            "epoch: 16729, dloss = 4.1, gloss = -2289.6, total_loss = -226.9 (0.69)\n",
            "epoch: 16730, dloss = 1.0, gloss = -2246.1, total_loss = -224.1 (0.69)\n",
            "epoch: 16731, dloss = 0.9, gloss = -2232.6, total_loss = -222.8 (0.83)\n",
            "epoch: 16732, dloss = 2.9, gloss = -2268.2, total_loss = -225.4 (0.88)\n",
            "epoch: 16733, dloss = 1.9, gloss = -2308.8, total_loss = -229.9 (0.67)\n",
            "epoch: 16734, dloss = -0.0, gloss = -2312.4, total_loss = -231.3 (0.70)\n",
            "epoch: 16735, dloss = -2.1, gloss = -2342.5, total_loss = -235.3 (0.68)\n",
            "epoch: 16736, dloss = 5.5, gloss = -2394.9, total_loss = -236.8 (0.66)\n",
            "epoch: 16737, dloss = 2.8, gloss = -2380.2, total_loss = -236.6 (0.68)\n",
            "epoch: 16738, dloss = 2.7, gloss = -2287.1, total_loss = -227.3 (0.70)\n",
            "epoch: 16739, dloss = 1.6, gloss = -2352.1, total_loss = -234.4 (0.68)\n",
            "epoch: 16740, dloss = 0.7, gloss = -2320.3, total_loss = -231.7 (0.66)\n",
            "epoch: 16741, dloss = -0.3, gloss = -2275.2, total_loss = -227.7 (0.67)\n",
            "epoch: 16742, dloss = 6.0, gloss = -2326.0, total_loss = -229.6 (0.69)\n",
            "epoch: 16743, dloss = -1.5, gloss = -2287.7, total_loss = -229.5 (0.70)\n",
            "epoch: 16744, dloss = -1.6, gloss = -2314.3, total_loss = -232.2 (0.69)\n",
            "epoch: 16745, dloss = 1.7, gloss = -2341.0, total_loss = -233.3 (0.68)\n",
            "epoch: 16746, dloss = 2.6, gloss = -2309.5, total_loss = -229.6 (0.69)\n",
            "epoch: 16747, dloss = -1.3, gloss = -2301.3, total_loss = -230.8 (0.92)\n",
            "epoch: 16748, dloss = -2.4, gloss = -2241.3, total_loss = -225.3 (0.79)\n",
            "epoch: 16749, dloss = 2.6, gloss = -2342.2, total_loss = -232.9 (0.67)\n",
            "epoch: 16750, dloss = 2.2, gloss = -2244.2, total_loss = -223.3 (0.68)\n",
            "epoch: 16751, dloss = 3.4, gloss = -2299.7, total_loss = -228.3 (0.70)\n",
            "epoch: 16752, dloss = -2.0, gloss = -2313.3, total_loss = -232.3 (0.67)\n",
            "epoch: 16753, dloss = 1.7, gloss = -2345.8, total_loss = -233.7 (0.69)\n",
            "epoch: 16754, dloss = 1.3, gloss = -2348.5, total_loss = -234.2 (0.68)\n",
            "epoch: 16755, dloss = 4.9, gloss = -2434.0, total_loss = -240.9 (0.67)\n",
            "epoch: 16756, dloss = 1.4, gloss = -2363.4, total_loss = -235.6 (0.67)\n",
            "epoch: 16757, dloss = 2.7, gloss = -2303.1, total_loss = -229.0 (0.69)\n",
            "epoch: 16758, dloss = 0.9, gloss = -2421.7, total_loss = -241.7 (0.67)\n",
            "epoch: 16759, dloss = -1.7, gloss = -2373.4, total_loss = -238.2 (0.66)\n",
            "epoch: 16760, dloss = 3.0, gloss = -2364.6, total_loss = -235.0 (0.67)\n",
            "epoch: 16761, dloss = 2.4, gloss = -2351.4, total_loss = -233.9 (0.69)\n",
            "epoch: 16762, dloss = -0.2, gloss = -2365.5, total_loss = -236.6 (0.70)\n",
            "epoch: 16763, dloss = 2.0, gloss = -2379.8, total_loss = -237.0 (1.02)\n",
            "epoch: 16764, dloss = -0.5, gloss = -2346.0, total_loss = -234.8 (0.67)\n",
            "epoch: 16765, dloss = 3.4, gloss = -2416.7, total_loss = -240.0 (0.66)\n",
            "epoch: 16766, dloss = 3.5, gloss = -2281.9, total_loss = -226.4 (0.70)\n",
            "epoch: 16767, dloss = 4.5, gloss = -2286.5, total_loss = -226.4 (0.70)\n",
            "epoch: 16768, dloss = 1.1, gloss = -2311.4, total_loss = -230.6 (0.68)\n",
            "epoch: 16769, dloss = 3.5, gloss = -2358.3, total_loss = -234.1 (0.68)\n",
            "epoch: 16770, dloss = 1.4, gloss = -2307.0, total_loss = -230.0 (0.69)\n",
            "epoch: 16771, dloss = -1.4, gloss = -2382.3, total_loss = -238.9 (0.68)\n",
            "epoch: 16772, dloss = -0.9, gloss = -2353.0, total_loss = -235.8 (0.68)\n",
            "epoch: 16773, dloss = 1.6, gloss = -2376.2, total_loss = -236.8 (0.67)\n",
            "epoch: 16774, dloss = 3.1, gloss = -2443.2, total_loss = -242.8 (0.67)\n",
            "epoch: 16775, dloss = 3.3, gloss = -2376.3, total_loss = -236.0 (0.69)\n",
            "epoch: 16776, dloss = -0.6, gloss = -2390.5, total_loss = -239.4 (0.67)\n",
            "epoch: 16777, dloss = -0.7, gloss = -2408.1, total_loss = -241.2 (0.67)\n",
            "epoch: 16778, dloss = -7.4, gloss = -2400.2, total_loss = -243.7 (0.79)\n",
            "epoch: 16779, dloss = -3.1, gloss = -2411.3, total_loss = -242.7 (0.93)\n",
            "epoch: 16780, dloss = -3.2, gloss = -2420.7, total_loss = -243.7 (0.69)\n",
            "epoch: 16781, dloss = -1.0, gloss = -2366.1, total_loss = -237.1 (0.69)\n",
            "epoch: 16782, dloss = 2.4, gloss = -2387.0, total_loss = -237.5 (0.68)\n",
            "epoch: 16783, dloss = 0.4, gloss = -2448.2, total_loss = -244.6 (0.69)\n",
            "epoch: 16784, dloss = 6.6, gloss = -2416.5, total_loss = -238.3 (0.70)\n",
            "epoch: 16785, dloss = 0.9, gloss = -2407.2, total_loss = -240.2 (0.70)\n",
            "epoch: 16786, dloss = -0.4, gloss = -2378.2, total_loss = -238.0 (0.68)\n",
            "epoch: 16787, dloss = -2.9, gloss = -2408.2, total_loss = -242.2 (0.68)\n",
            "epoch: 16788, dloss = 2.9, gloss = -2460.8, total_loss = -244.6 (0.68)\n",
            "epoch: 16789, dloss = 0.0, gloss = -2488.9, total_loss = -248.9 (0.70)\n",
            "epoch: 16790, dloss = -1.4, gloss = -2468.5, total_loss = -247.5 (0.67)\n",
            "epoch: 16791, dloss = 2.9, gloss = -2476.6, total_loss = -246.2 (0.67)\n",
            "epoch: 16792, dloss = 1.3, gloss = -2420.5, total_loss = -241.4 (0.69)\n",
            "epoch: 16793, dloss = 3.1, gloss = -2390.2, total_loss = -237.5 (0.67)\n",
            "epoch: 16794, dloss = 2.2, gloss = -2374.7, total_loss = -236.4 (0.91)\n",
            "epoch: 16795, dloss = 1.3, gloss = -2385.1, total_loss = -237.8 (0.82)\n",
            "epoch: 16796, dloss = -1.9, gloss = -2331.1, total_loss = -234.0 (0.68)\n",
            "epoch: 16797, dloss = 2.3, gloss = -2478.4, total_loss = -246.7 (0.66)\n",
            "epoch: 16798, dloss = 3.4, gloss = -2447.5, total_loss = -243.1 (0.70)\n",
            "epoch: 16799, dloss = 1.9, gloss = -2485.7, total_loss = -247.6 (0.66)\n",
            "epoch: 16800, dloss = 1.3, gloss = -2485.6, total_loss = -247.9 (0.67)\n",
            "epoch: 16801, dloss = -3.8, gloss = -2409.4, total_loss = -242.9 (0.69)\n",
            "epoch: 16802, dloss = 1.9, gloss = -2372.1, total_loss = -236.3 (0.69)\n",
            "epoch: 16803, dloss = -0.7, gloss = -2410.2, total_loss = -241.4 (0.66)\n",
            "epoch: 16804, dloss = 2.0, gloss = -2437.9, total_loss = -242.8 (0.67)\n",
            "epoch: 16805, dloss = -0.2, gloss = -2349.7, total_loss = -235.1 (0.70)\n",
            "epoch: 16806, dloss = 2.6, gloss = -2341.0, total_loss = -232.8 (0.71)\n",
            "epoch: 16807, dloss = 0.1, gloss = -2391.6, total_loss = -239.1 (0.67)\n",
            "epoch: 16808, dloss = -2.1, gloss = -2468.0, total_loss = -247.8 (0.70)\n",
            "epoch: 16809, dloss = 2.3, gloss = -2351.9, total_loss = -234.0 (0.96)\n",
            "epoch: 16810, dloss = -1.3, gloss = -2368.3, total_loss = -237.5 (1.14)\n",
            "epoch: 16811, dloss = 3.2, gloss = -2342.2, total_loss = -232.6 (0.72)\n",
            "epoch: 16812, dloss = 1.4, gloss = -2394.4, total_loss = -238.7 (0.67)\n",
            "epoch: 16813, dloss = 0.5, gloss = -2303.1, total_loss = -230.0 (0.65)\n",
            "epoch: 16814, dloss = -0.6, gloss = -2337.9, total_loss = -234.1 (0.66)\n",
            "epoch: 16815, dloss = -0.6, gloss = -2376.5, total_loss = -237.9 (0.67)\n",
            "epoch: 16816, dloss = -1.6, gloss = -2361.8, total_loss = -237.0 (0.68)\n",
            "epoch: 16817, dloss = 3.0, gloss = -2339.7, total_loss = -232.5 (0.68)\n",
            "epoch: 16818, dloss = 4.7, gloss = -2313.0, total_loss = -228.9 (0.70)\n",
            "epoch: 16819, dloss = -1.3, gloss = -2351.1, total_loss = -235.7 (0.92)\n",
            "epoch: 16820, dloss = 4.1, gloss = -2336.9, total_loss = -231.7 (0.84)\n",
            "epoch: 16821, dloss = -0.8, gloss = -2340.4, total_loss = -234.5 (0.66)\n",
            "epoch: 16822, dloss = -0.1, gloss = -2401.6, total_loss = -240.2 (0.69)\n",
            "epoch: 16823, dloss = 2.0, gloss = -2346.1, total_loss = -233.6 (0.68)\n",
            "epoch: 16824, dloss = -3.0, gloss = -2361.5, total_loss = -237.6 (0.69)\n",
            "epoch: 16825, dloss = -1.3, gloss = -2266.4, total_loss = -227.3 (0.95)\n",
            "epoch: 16826, dloss = -2.0, gloss = -2294.8, total_loss = -230.5 (0.82)\n",
            "epoch: 16827, dloss = -1.5, gloss = -2297.5, total_loss = -230.5 (0.69)\n",
            "epoch: 16828, dloss = 0.9, gloss = -2242.6, total_loss = -223.8 (0.70)\n",
            "epoch: 16829, dloss = 2.7, gloss = -2441.8, total_loss = -242.8 (0.68)\n",
            "epoch: 16830, dloss = 1.3, gloss = -2320.0, total_loss = -231.4 (0.67)\n",
            "epoch: 16831, dloss = 1.6, gloss = -2364.7, total_loss = -235.7 (0.68)\n",
            "epoch: 16832, dloss = 3.4, gloss = -2331.0, total_loss = -231.4 (0.68)\n",
            "epoch: 16833, dloss = 3.3, gloss = -2369.7, total_loss = -235.3 (0.70)\n",
            "epoch: 16834, dloss = 1.6, gloss = -2334.0, total_loss = -232.6 (0.67)\n",
            "epoch: 16835, dloss = 2.7, gloss = -2344.0, total_loss = -233.0 (0.70)\n",
            "epoch: 16836, dloss = -2.4, gloss = -2320.4, total_loss = -233.2 (0.71)\n",
            "epoch: 16837, dloss = 0.9, gloss = -2418.7, total_loss = -241.4 (0.70)\n",
            "epoch: 16838, dloss = 1.6, gloss = -2422.7, total_loss = -241.5 (0.67)\n",
            "epoch: 16839, dloss = 1.3, gloss = -2403.3, total_loss = -239.7 (0.71)\n",
            "epoch: 16840, dloss = 0.5, gloss = -2336.7, total_loss = -233.4 (0.79)\n",
            "epoch: 16841, dloss = 0.2, gloss = -2283.5, total_loss = -228.2 (0.96)\n",
            "epoch: 16842, dloss = -0.1, gloss = -2288.8, total_loss = -228.9 (0.70)\n",
            "epoch: 16843, dloss = 2.1, gloss = -2360.3, total_loss = -235.0 (0.70)\n",
            "epoch: 16844, dloss = 1.4, gloss = -2286.7, total_loss = -228.0 (0.70)\n",
            "epoch: 16845, dloss = -1.6, gloss = -2343.3, total_loss = -235.1 (0.69)\n",
            "epoch: 16846, dloss = 1.9, gloss = -2350.9, total_loss = -234.1 (0.68)\n",
            "epoch: 16847, dloss = -0.3, gloss = -2349.4, total_loss = -235.1 (0.69)\n",
            "epoch: 16848, dloss = -0.5, gloss = -2308.4, total_loss = -231.1 (0.68)\n",
            "epoch: 16849, dloss = 2.7, gloss = -2404.8, total_loss = -239.2 (0.68)\n",
            "epoch: 16850, dloss = 2.2, gloss = -2451.5, total_loss = -244.0 (0.70)\n",
            "epoch: 16851, dloss = 1.8, gloss = -2351.1, total_loss = -234.2 (0.67)\n",
            "epoch: 16852, dloss = -3.7, gloss = -2379.1, total_loss = -239.7 (0.68)\n",
            "epoch: 16853, dloss = 0.2, gloss = -2380.0, total_loss = -237.9 (0.72)\n",
            "epoch: 16854, dloss = 7.4, gloss = -2368.0, total_loss = -233.1 (0.68)\n",
            "epoch: 16855, dloss = 1.4, gloss = -2375.2, total_loss = -236.8 (0.69)\n",
            "epoch: 16856, dloss = 3.6, gloss = -2383.5, total_loss = -236.5 (0.93)\n",
            "epoch: 16857, dloss = 1.4, gloss = -2352.3, total_loss = -234.5 (0.80)\n",
            "epoch: 16858, dloss = 2.8, gloss = -2418.3, total_loss = -240.4 (0.66)\n",
            "epoch: 16859, dloss = 2.5, gloss = -2377.5, total_loss = -236.5 (0.67)\n",
            "epoch: 16860, dloss = -0.7, gloss = -2411.0, total_loss = -241.5 (0.69)\n",
            "epoch: 16861, dloss = -0.3, gloss = -2391.4, total_loss = -239.3 (0.67)\n",
            "epoch: 16862, dloss = 0.0, gloss = -2352.5, total_loss = -235.2 (0.70)\n",
            "epoch: 16863, dloss = 0.7, gloss = -2391.3, total_loss = -238.8 (0.68)\n",
            "epoch: 16864, dloss = -4.4, gloss = -2247.7, total_loss = -227.0 (0.68)\n",
            "epoch: 16865, dloss = 0.9, gloss = -2304.0, total_loss = -229.9 (0.68)\n",
            "epoch: 16866, dloss = -5.2, gloss = -2309.6, total_loss = -233.6 (0.68)\n",
            "epoch: 16867, dloss = 1.2, gloss = -2367.2, total_loss = -236.1 (0.67)\n",
            "epoch: 16868, dloss = -2.2, gloss = -2430.2, total_loss = -244.1 (0.68)\n",
            "epoch: 16869, dloss = -0.1, gloss = -2432.7, total_loss = -243.3 (0.68)\n",
            "epoch: 16870, dloss = 2.7, gloss = -2350.6, total_loss = -233.7 (0.69)\n",
            "epoch: 16871, dloss = 0.9, gloss = -2308.0, total_loss = -230.3 (0.66)\n",
            "epoch: 16872, dloss = -1.4, gloss = -2379.1, total_loss = -238.6 (1.27)\n",
            "epoch: 16873, dloss = 3.8, gloss = -2355.0, total_loss = -233.6 (0.68)\n",
            "epoch: 16874, dloss = 0.1, gloss = -2365.5, total_loss = -236.5 (0.70)\n",
            "epoch: 16875, dloss = 2.3, gloss = -2404.8, total_loss = -239.3 (0.68)\n",
            "epoch: 16876, dloss = -0.0, gloss = -2383.1, total_loss = -238.3 (0.66)\n",
            "epoch: 16877, dloss = 0.9, gloss = -2410.4, total_loss = -240.6 (0.70)\n",
            "epoch: 16878, dloss = 2.5, gloss = -2410.6, total_loss = -239.8 (0.66)\n",
            "epoch: 16879, dloss = 0.3, gloss = -2423.2, total_loss = -242.2 (0.69)\n",
            "epoch: 16880, dloss = -0.8, gloss = -2352.1, total_loss = -235.6 (0.69)\n",
            "epoch: 16881, dloss = -1.1, gloss = -2273.2, total_loss = -227.9 (0.68)\n",
            "epoch: 16882, dloss = 1.2, gloss = -2314.0, total_loss = -230.8 (0.68)\n",
            "epoch: 16883, dloss = 2.3, gloss = -2328.8, total_loss = -231.7 (0.68)\n",
            "epoch: 16884, dloss = 8.3, gloss = -2377.2, total_loss = -233.6 (0.67)\n",
            "epoch: 16885, dloss = 2.7, gloss = -2387.9, total_loss = -237.5 (0.68)\n",
            "epoch: 16886, dloss = 0.0, gloss = -2345.2, total_loss = -234.5 (0.68)\n",
            "epoch: 16887, dloss = 0.6, gloss = -2325.0, total_loss = -232.2 (0.90)\n",
            "epoch: 16888, dloss = 3.1, gloss = -2341.6, total_loss = -232.6 (0.84)\n",
            "epoch: 16889, dloss = 38.3, gloss = -2283.4, total_loss = -209.2 (0.68)\n",
            "epoch: 16890, dloss = -0.9, gloss = -2312.1, total_loss = -231.7 (0.66)\n",
            "epoch: 16891, dloss = -0.2, gloss = -2347.0, total_loss = -234.8 (0.68)\n",
            "epoch: 16892, dloss = 1.0, gloss = -2348.0, total_loss = -234.3 (0.69)\n",
            "epoch: 16893, dloss = 0.3, gloss = -2312.6, total_loss = -231.1 (0.69)\n",
            "epoch: 16894, dloss = 0.5, gloss = -2281.2, total_loss = -227.8 (0.69)\n",
            "epoch: 16895, dloss = 5.3, gloss = -2375.4, total_loss = -234.9 (0.70)\n",
            "epoch: 16896, dloss = 3.2, gloss = -2301.1, total_loss = -228.5 (0.68)\n",
            "epoch: 16897, dloss = -0.9, gloss = -2346.9, total_loss = -235.1 (0.66)\n",
            "epoch: 16898, dloss = 5.2, gloss = -2414.6, total_loss = -238.9 (0.68)\n",
            "epoch: 16899, dloss = -0.3, gloss = -2361.3, total_loss = -236.3 (0.66)\n",
            "epoch: 16900, dloss = -1.2, gloss = -2284.4, total_loss = -229.0 (0.68)\n",
            "epoch: 16901, dloss = -1.1, gloss = -2373.6, total_loss = -237.9 (0.70)\n",
            "epoch: 16902, dloss = 0.4, gloss = -2282.6, total_loss = -228.1 (0.67)\n",
            "epoch: 16903, dloss = 0.5, gloss = -2299.8, total_loss = -229.7 (0.96)\n",
            "epoch: 16904, dloss = -1.3, gloss = -2283.0, total_loss = -228.9 (0.74)\n",
            "epoch: 16905, dloss = 1.7, gloss = -2367.4, total_loss = -235.9 (0.68)\n",
            "epoch: 16906, dloss = 0.7, gloss = -2373.0, total_loss = -236.9 (0.69)\n",
            "epoch: 16907, dloss = -3.3, gloss = -2348.6, total_loss = -236.5 (0.69)\n",
            "epoch: 16908, dloss = 1.4, gloss = -2328.8, total_loss = -232.2 (0.69)\n",
            "epoch: 16909, dloss = 0.2, gloss = -2358.5, total_loss = -235.7 (0.70)\n",
            "epoch: 16910, dloss = 2.8, gloss = -2341.8, total_loss = -232.8 (0.69)\n",
            "epoch: 16911, dloss = 2.6, gloss = -2404.0, total_loss = -239.1 (0.68)\n",
            "epoch: 16912, dloss = -2.6, gloss = -2370.9, total_loss = -238.4 (0.68)\n",
            "epoch: 16913, dloss = 1.7, gloss = -2391.0, total_loss = -238.2 (0.70)\n",
            "epoch: 16914, dloss = 2.8, gloss = -2389.5, total_loss = -237.5 (0.67)\n",
            "epoch: 16915, dloss = 3.8, gloss = -2375.1, total_loss = -235.6 (0.68)\n",
            "epoch: 16916, dloss = 1.0, gloss = -2347.4, total_loss = -234.2 (0.71)\n",
            "epoch: 16917, dloss = 4.1, gloss = -2347.6, total_loss = -232.7 (0.68)\n",
            "epoch: 16918, dloss = 3.9, gloss = -2362.8, total_loss = -234.3 (0.79)\n",
            "epoch: 16919, dloss = 1.9, gloss = -2330.9, total_loss = -232.2 (0.92)\n",
            "epoch: 16920, dloss = -0.2, gloss = -2236.2, total_loss = -223.7 (0.68)\n",
            "epoch: 16921, dloss = 1.0, gloss = -2300.2, total_loss = -229.5 (0.67)\n",
            "epoch: 16922, dloss = -0.2, gloss = -2373.5, total_loss = -237.5 (0.68)\n",
            "epoch: 16923, dloss = 0.5, gloss = -2436.7, total_loss = -243.4 (0.67)\n",
            "epoch: 16924, dloss = -2.1, gloss = -2278.3, total_loss = -228.9 (0.68)\n",
            "epoch: 16925, dloss = -0.3, gloss = -2313.5, total_loss = -231.5 (0.70)\n",
            "epoch: 16926, dloss = 1.0, gloss = -2333.3, total_loss = -232.8 (0.70)\n",
            "epoch: 16927, dloss = -1.3, gloss = -2310.4, total_loss = -231.7 (0.68)\n",
            "epoch: 16928, dloss = 1.9, gloss = -2339.3, total_loss = -233.0 (0.72)\n",
            "epoch: 16929, dloss = -1.4, gloss = -2326.6, total_loss = -233.4 (0.68)\n",
            "epoch: 16930, dloss = 0.4, gloss = -2447.1, total_loss = -244.5 (0.70)\n",
            "epoch: 16931, dloss = 1.4, gloss = -2459.0, total_loss = -245.2 (0.68)\n",
            "epoch: 16932, dloss = 4.4, gloss = -2466.4, total_loss = -244.4 (0.70)\n",
            "epoch: 16933, dloss = 0.8, gloss = -2451.7, total_loss = -244.8 (0.69)\n",
            "epoch: 16934, dloss = 0.7, gloss = -2413.0, total_loss = -241.0 (1.26)\n",
            "epoch: 16935, dloss = 2.6, gloss = -2407.1, total_loss = -239.4 (0.70)\n",
            "epoch: 16936, dloss = 2.2, gloss = -2356.2, total_loss = -234.5 (0.69)\n",
            "epoch: 16937, dloss = 0.6, gloss = -2417.5, total_loss = -241.4 (0.69)\n",
            "epoch: 16938, dloss = 2.3, gloss = -2401.2, total_loss = -239.0 (0.69)\n",
            "epoch: 16939, dloss = -0.0, gloss = -2322.2, total_loss = -232.2 (0.70)\n",
            "epoch: 16940, dloss = 2.8, gloss = -2382.4, total_loss = -236.8 (0.70)\n",
            "epoch: 16941, dloss = 3.4, gloss = -2435.7, total_loss = -241.9 (0.69)\n",
            "epoch: 16942, dloss = 3.2, gloss = -2487.8, total_loss = -247.2 (0.68)\n",
            "epoch: 16943, dloss = 3.0, gloss = -2448.9, total_loss = -243.4 (0.69)\n",
            "epoch: 16944, dloss = 2.9, gloss = -2499.0, total_loss = -248.4 (0.69)\n",
            "epoch: 16945, dloss = 4.5, gloss = -2393.8, total_loss = -237.1 (0.69)\n",
            "epoch: 16946, dloss = 3.3, gloss = -2478.6, total_loss = -246.2 (0.68)\n",
            "epoch: 16947, dloss = 4.3, gloss = -2506.5, total_loss = -248.5 (0.67)\n",
            "epoch: 16948, dloss = 1.6, gloss = -2560.6, total_loss = -255.3 (0.68)\n",
            "epoch: 16949, dloss = 1.7, gloss = -2465.3, total_loss = -245.7 (0.88)\n",
            "epoch: 16950, dloss = 1.8, gloss = -2420.5, total_loss = -241.2 (0.83)\n",
            "epoch: 16951, dloss = 4.8, gloss = -2496.0, total_loss = -247.2 (0.66)\n",
            "epoch: 16952, dloss = 0.4, gloss = -2383.0, total_loss = -238.1 (0.68)\n",
            "epoch: 16953, dloss = -0.4, gloss = -2402.8, total_loss = -240.5 (0.68)\n",
            "epoch: 16954, dloss = 1.9, gloss = -2331.9, total_loss = -232.2 (0.67)\n",
            "epoch: 16955, dloss = 2.2, gloss = -2396.1, total_loss = -238.5 (0.68)\n",
            "epoch: 16956, dloss = -0.8, gloss = -2306.6, total_loss = -231.1 (0.68)\n",
            "epoch: 16957, dloss = -1.2, gloss = -2380.7, total_loss = -238.7 (0.69)\n",
            "epoch: 16958, dloss = -0.6, gloss = -2365.7, total_loss = -236.9 (0.68)\n",
            "epoch: 16959, dloss = -2.0, gloss = -2426.9, total_loss = -243.7 (0.72)\n",
            "epoch: 16960, dloss = 4.3, gloss = -2474.2, total_loss = -245.3 (0.69)\n",
            "epoch: 16961, dloss = 2.8, gloss = -2516.4, total_loss = -250.2 (0.67)\n",
            "epoch: 16962, dloss = 1.3, gloss = -2503.6, total_loss = -249.7 (0.70)\n",
            "epoch: 16963, dloss = 3.8, gloss = -2479.6, total_loss = -246.1 (0.66)\n",
            "epoch: 16964, dloss = 3.4, gloss = -2359.6, total_loss = -234.3 (0.68)\n",
            "epoch: 16965, dloss = 0.2, gloss = -2347.1, total_loss = -234.6 (1.26)\n",
            "epoch: 16966, dloss = 0.1, gloss = -2377.8, total_loss = -237.7 (0.69)\n",
            "epoch: 16967, dloss = 0.1, gloss = -2414.1, total_loss = -241.4 (0.69)\n",
            "epoch: 16968, dloss = 1.4, gloss = -2419.5, total_loss = -241.2 (0.67)\n",
            "epoch: 16969, dloss = 1.5, gloss = -2459.1, total_loss = -245.2 (0.70)\n",
            "epoch: 16970, dloss = 2.3, gloss = -2337.7, total_loss = -232.6 (0.69)\n",
            "epoch: 16971, dloss = -1.3, gloss = -2461.8, total_loss = -246.9 (0.68)\n",
            "epoch: 16972, dloss = 1.9, gloss = -2422.7, total_loss = -241.3 (0.68)\n",
            "epoch: 16973, dloss = -0.8, gloss = -2424.6, total_loss = -242.8 (0.70)\n",
            "epoch: 16974, dloss = 1.7, gloss = -2466.3, total_loss = -245.8 (0.91)\n",
            "epoch: 16975, dloss = 1.6, gloss = -2428.2, total_loss = -242.0 (0.95)\n",
            "epoch: 16976, dloss = -2.7, gloss = -2358.3, total_loss = -237.2 (0.69)\n",
            "epoch: 16977, dloss = 0.7, gloss = -2370.8, total_loss = -236.7 (0.68)\n",
            "epoch: 16978, dloss = 3.3, gloss = -2395.0, total_loss = -237.9 (0.69)\n",
            "epoch: 16979, dloss = 1.0, gloss = -2469.4, total_loss = -246.4 (0.85)\n",
            "epoch: 16980, dloss = 0.2, gloss = -2388.4, total_loss = -238.7 (0.89)\n",
            "epoch: 16981, dloss = -0.5, gloss = -2342.9, total_loss = -234.5 (0.69)\n",
            "epoch: 16982, dloss = 3.1, gloss = -2438.6, total_loss = -242.3 (0.68)\n",
            "epoch: 16983, dloss = -0.8, gloss = -2363.7, total_loss = -236.7 (0.66)\n",
            "epoch: 16984, dloss = 2.7, gloss = -2393.0, total_loss = -238.0 (0.68)\n",
            "epoch: 16985, dloss = -0.3, gloss = -2397.2, total_loss = -239.9 (0.68)\n",
            "epoch: 16986, dloss = 3.1, gloss = -2479.9, total_loss = -246.4 (0.69)\n",
            "epoch: 16987, dloss = -0.7, gloss = -2469.4, total_loss = -247.3 (0.69)\n",
            "epoch: 16988, dloss = 3.9, gloss = -2523.3, total_loss = -250.4 (0.68)\n",
            "epoch: 16989, dloss = -0.4, gloss = -2437.2, total_loss = -243.9 (0.68)\n",
            "epoch: 16990, dloss = -0.1, gloss = -2393.4, total_loss = -239.4 (0.70)\n",
            "epoch: 16991, dloss = 0.3, gloss = -2324.5, total_loss = -232.3 (0.69)\n",
            "epoch: 16992, dloss = 0.3, gloss = -2394.8, total_loss = -239.3 (0.69)\n",
            "epoch: 16993, dloss = 1.2, gloss = -2396.6, total_loss = -239.1 (0.68)\n",
            "epoch: 16994, dloss = 3.0, gloss = -2471.1, total_loss = -245.6 (0.69)\n",
            "epoch: 16995, dloss = 0.1, gloss = -2397.2, total_loss = -239.7 (0.99)\n",
            "epoch: 16996, dloss = -0.4, gloss = -2390.4, total_loss = -239.2 (0.78)\n",
            "epoch: 16997, dloss = 2.6, gloss = -2396.4, total_loss = -238.3 (0.72)\n",
            "epoch: 16998, dloss = 2.1, gloss = -2395.2, total_loss = -238.5 (0.70)\n",
            "epoch: 16999, dloss = -3.2, gloss = -2404.3, total_loss = -242.0 (0.68)\n",
            "epoch: 17000, dloss = 1.2, gloss = -2456.6, total_loss = -245.0 (0.71)\n",
            "epoch: 17001, dloss = 1.3, gloss = -2470.4, total_loss = -246.4 (0.67)\n",
            "epoch: 17002, dloss = 0.9, gloss = -2408.8, total_loss = -240.4 (0.69)\n",
            "epoch: 17003, dloss = 1.1, gloss = -2348.7, total_loss = -234.3 (0.69)\n",
            "epoch: 17004, dloss = -0.1, gloss = -2395.2, total_loss = -239.6 (0.69)\n",
            "epoch: 17005, dloss = -0.8, gloss = -2348.8, total_loss = -235.3 (0.67)\n",
            "epoch: 17006, dloss = 2.8, gloss = -2432.2, total_loss = -241.8 (0.69)\n",
            "epoch: 17007, dloss = 0.5, gloss = -2408.4, total_loss = -240.6 (0.68)\n",
            "epoch: 17008, dloss = 0.4, gloss = -2451.9, total_loss = -245.0 (0.66)\n",
            "epoch: 17009, dloss = -0.7, gloss = -2437.0, total_loss = -244.1 (0.68)\n",
            "epoch: 17010, dloss = 1.8, gloss = -2445.4, total_loss = -243.7 (0.83)\n",
            "epoch: 17011, dloss = 2.8, gloss = -2449.7, total_loss = -243.6 (0.93)\n",
            "epoch: 17012, dloss = 2.3, gloss = -2464.4, total_loss = -245.3 (0.71)\n",
            "epoch: 17013, dloss = -1.8, gloss = -2397.0, total_loss = -240.6 (0.68)\n",
            "epoch: 17014, dloss = 2.5, gloss = -2350.6, total_loss = -233.8 (0.72)\n",
            "epoch: 17015, dloss = 0.0, gloss = -2382.2, total_loss = -238.2 (0.69)\n",
            "epoch: 17016, dloss = 1.8, gloss = -2392.5, total_loss = -238.3 (0.70)\n",
            "epoch: 17017, dloss = 0.9, gloss = -2327.8, total_loss = -232.3 (0.69)\n",
            "epoch: 17018, dloss = 2.2, gloss = -2377.1, total_loss = -236.6 (0.71)\n",
            "epoch: 17019, dloss = 1.1, gloss = -2389.2, total_loss = -238.4 (0.68)\n",
            "epoch: 17020, dloss = 3.6, gloss = -2372.9, total_loss = -235.5 (0.69)\n",
            "epoch: 17021, dloss = -1.6, gloss = -2354.0, total_loss = -236.2 (0.70)\n",
            "epoch: 17022, dloss = 3.8, gloss = -2354.7, total_loss = -233.6 (0.66)\n",
            "epoch: 17023, dloss = 2.9, gloss = -2394.8, total_loss = -238.0 (0.66)\n",
            "epoch: 17024, dloss = 3.0, gloss = -2383.1, total_loss = -236.8 (0.69)\n",
            "epoch: 17025, dloss = 2.1, gloss = -2390.2, total_loss = -238.0 (0.66)\n",
            "epoch: 17026, dloss = -0.6, gloss = -2404.1, total_loss = -240.7 (1.03)\n",
            "epoch: 17027, dloss = -1.8, gloss = -2297.6, total_loss = -230.7 (0.74)\n",
            "epoch: 17028, dloss = 0.8, gloss = -2368.1, total_loss = -236.4 (0.70)\n",
            "epoch: 17029, dloss = -0.3, gloss = -2359.5, total_loss = -236.1 (0.68)\n",
            "epoch: 17030, dloss = 1.6, gloss = -2416.2, total_loss = -240.8 (0.72)\n",
            "epoch: 17031, dloss = 0.2, gloss = -2415.2, total_loss = -241.4 (0.66)\n",
            "epoch: 17032, dloss = 0.7, gloss = -2418.4, total_loss = -241.5 (0.71)\n",
            "epoch: 17033, dloss = -1.3, gloss = -2534.5, total_loss = -254.1 (0.71)\n",
            "epoch: 17034, dloss = 3.3, gloss = -2477.0, total_loss = -246.1 (0.71)\n",
            "epoch: 17035, dloss = 3.0, gloss = -2467.1, total_loss = -245.2 (0.68)\n",
            "epoch: 17036, dloss = 1.3, gloss = -2415.7, total_loss = -240.9 (0.68)\n",
            "epoch: 17037, dloss = -4.5, gloss = -2434.5, total_loss = -245.7 (0.69)\n",
            "epoch: 17038, dloss = -0.5, gloss = -2507.4, total_loss = -251.0 (0.68)\n",
            "epoch: 17039, dloss = -0.5, gloss = -2437.9, total_loss = -244.1 (0.70)\n",
            "epoch: 17040, dloss = 1.2, gloss = -2441.8, total_loss = -243.6 (0.73)\n",
            "epoch: 17041, dloss = 4.4, gloss = -2428.2, total_loss = -240.6 (0.89)\n",
            "epoch: 17042, dloss = 1.6, gloss = -2412.2, total_loss = -240.4 (0.89)\n",
            "epoch: 17043, dloss = 1.8, gloss = -2379.4, total_loss = -237.0 (0.69)\n",
            "epoch: 17044, dloss = 5.9, gloss = -2398.9, total_loss = -236.9 (0.69)\n",
            "epoch: 17045, dloss = 1.7, gloss = -2387.7, total_loss = -237.9 (0.69)\n",
            "epoch: 17046, dloss = 2.6, gloss = -2543.5, total_loss = -253.0 (0.69)\n",
            "epoch: 17047, dloss = 2.2, gloss = -2414.0, total_loss = -240.3 (0.68)\n",
            "epoch: 17048, dloss = 2.6, gloss = -2482.0, total_loss = -246.9 (0.70)\n",
            "epoch: 17049, dloss = 1.6, gloss = -2461.7, total_loss = -245.4 (0.68)\n",
            "epoch: 17050, dloss = 0.4, gloss = -2462.4, total_loss = -246.0 (0.70)\n",
            "epoch: 17051, dloss = 1.7, gloss = -2418.7, total_loss = -241.0 (0.70)\n",
            "epoch: 17052, dloss = 1.0, gloss = -2476.9, total_loss = -247.2 (0.70)\n",
            "epoch: 17053, dloss = 4.7, gloss = -2513.3, total_loss = -249.0 (0.69)\n",
            "epoch: 17054, dloss = 1.9, gloss = -2444.0, total_loss = -243.4 (0.66)\n",
            "epoch: 17055, dloss = 3.1, gloss = -2398.3, total_loss = -238.3 (0.68)\n",
            "epoch: 17056, dloss = 0.9, gloss = -2471.4, total_loss = -246.7 (0.70)\n",
            "epoch: 17057, dloss = -1.7, gloss = -2437.6, total_loss = -244.6 (1.04)\n",
            "epoch: 17058, dloss = -0.3, gloss = -2418.2, total_loss = -242.0 (0.72)\n",
            "epoch: 17059, dloss = 1.6, gloss = -2396.9, total_loss = -238.9 (0.87)\n",
            "epoch: 17060, dloss = 0.9, gloss = -2403.1, total_loss = -239.9 (0.84)\n",
            "epoch: 17061, dloss = 0.9, gloss = -2423.1, total_loss = -241.8 (0.69)\n",
            "epoch: 17062, dloss = 0.5, gloss = -2442.3, total_loss = -244.0 (0.69)\n",
            "epoch: 17063, dloss = -1.6, gloss = -2423.9, total_loss = -243.2 (0.68)\n",
            "epoch: 17064, dloss = 2.5, gloss = -2430.3, total_loss = -241.8 (0.69)\n",
            "epoch: 17065, dloss = 2.9, gloss = -2407.9, total_loss = -239.3 (0.68)\n",
            "epoch: 17066, dloss = 1.2, gloss = -2434.0, total_loss = -242.8 (0.70)\n",
            "epoch: 17067, dloss = 0.4, gloss = -2425.2, total_loss = -242.3 (0.73)\n",
            "epoch: 17068, dloss = 1.1, gloss = -2442.9, total_loss = -243.7 (1.01)\n",
            "epoch: 17069, dloss = 2.2, gloss = -2419.7, total_loss = -240.9 (0.67)\n",
            "epoch: 17070, dloss = 0.8, gloss = -2334.0, total_loss = -233.0 (0.70)\n",
            "epoch: 17071, dloss = -3.0, gloss = -2337.2, total_loss = -235.2 (0.86)\n",
            "epoch: 17072, dloss = 2.5, gloss = -2380.5, total_loss = -236.8 (0.85)\n",
            "epoch: 17073, dloss = -2.8, gloss = -2338.0, total_loss = -235.2 (0.70)\n",
            "epoch: 17074, dloss = 0.2, gloss = -2453.4, total_loss = -245.2 (0.71)\n",
            "epoch: 17075, dloss = 1.0, gloss = -2268.1, total_loss = -226.3 (0.68)\n",
            "epoch: 17076, dloss = 2.1, gloss = -2301.8, total_loss = -229.1 (0.68)\n",
            "epoch: 17077, dloss = 2.0, gloss = -2279.0, total_loss = -226.9 (0.68)\n",
            "epoch: 17078, dloss = 2.5, gloss = -2365.0, total_loss = -235.3 (0.67)\n",
            "epoch: 17079, dloss = -0.3, gloss = -2385.7, total_loss = -238.7 (0.68)\n",
            "epoch: 17080, dloss = 0.5, gloss = -2361.1, total_loss = -235.9 (0.66)\n",
            "epoch: 17081, dloss = 0.8, gloss = -2436.3, total_loss = -243.2 (0.68)\n",
            "epoch: 17082, dloss = 1.4, gloss = -2396.7, total_loss = -239.0 (0.70)\n",
            "epoch: 17083, dloss = 0.5, gloss = -2354.0, total_loss = -235.2 (0.69)\n",
            "epoch: 17084, dloss = 0.9, gloss = -2425.1, total_loss = -242.1 (0.67)\n",
            "epoch: 17085, dloss = 0.4, gloss = -2344.5, total_loss = -234.3 (0.71)\n",
            "epoch: 17086, dloss = 2.3, gloss = -2385.4, total_loss = -237.4 (0.71)\n",
            "epoch: 17087, dloss = 0.2, gloss = -2385.7, total_loss = -238.5 (1.03)\n",
            "epoch: 17088, dloss = 1.4, gloss = -2310.3, total_loss = -230.3 (0.67)\n",
            "epoch: 17089, dloss = 2.4, gloss = -2342.8, total_loss = -233.1 (0.66)\n",
            "epoch: 17090, dloss = 2.5, gloss = -2332.6, total_loss = -232.0 (0.67)\n",
            "epoch: 17091, dloss = 2.0, gloss = -2365.2, total_loss = -235.5 (0.67)\n",
            "epoch: 17092, dloss = -1.6, gloss = -2292.9, total_loss = -230.1 (0.67)\n",
            "epoch: 17093, dloss = 0.8, gloss = -2276.2, total_loss = -227.2 (0.70)\n",
            "epoch: 17094, dloss = 0.5, gloss = -2347.1, total_loss = -234.5 (0.68)\n",
            "epoch: 17095, dloss = -0.9, gloss = -2367.1, total_loss = -237.1 (0.69)\n",
            "epoch: 17096, dloss = -3.2, gloss = -2421.7, total_loss = -243.8 (0.68)\n",
            "epoch: 17097, dloss = 0.4, gloss = -2357.5, total_loss = -235.6 (0.67)\n",
            "epoch: 17098, dloss = -0.1, gloss = -2390.2, total_loss = -239.0 (0.70)\n",
            "epoch: 17099, dloss = 0.4, gloss = -2321.5, total_loss = -232.0 (0.68)\n",
            "epoch: 17100, dloss = 1.8, gloss = -2294.4, total_loss = -228.6 (0.67)\n",
            "epoch: 17101, dloss = 0.1, gloss = -2280.0, total_loss = -228.0 (0.68)\n",
            "epoch: 17102, dloss = -0.0, gloss = -2291.8, total_loss = -229.2 (0.77)\n",
            "epoch: 17103, dloss = 1.1, gloss = -2330.7, total_loss = -232.5 (0.95)\n",
            "epoch: 17104, dloss = 0.4, gloss = -2328.7, total_loss = -232.7 (0.67)\n",
            "epoch: 17105, dloss = 1.4, gloss = -2378.5, total_loss = -237.1 (0.68)\n",
            "epoch: 17106, dloss = 2.7, gloss = -2415.6, total_loss = -240.2 (0.69)\n",
            "epoch: 17107, dloss = 2.7, gloss = -2433.4, total_loss = -242.0 (0.68)\n",
            "epoch: 17108, dloss = -2.6, gloss = -2304.4, total_loss = -231.7 (0.68)\n",
            "epoch: 17109, dloss = -3.2, gloss = -2527.1, total_loss = -254.3 (0.68)\n",
            "epoch: 17110, dloss = -0.6, gloss = -2450.1, total_loss = -245.3 (0.68)\n",
            "epoch: 17111, dloss = -6.6, gloss = -2446.4, total_loss = -247.9 (0.67)\n",
            "epoch: 17112, dloss = 1.4, gloss = -2347.2, total_loss = -234.0 (0.69)\n",
            "epoch: 17113, dloss = 3.0, gloss = -2378.2, total_loss = -236.3 (0.68)\n",
            "epoch: 17114, dloss = 1.5, gloss = -2427.4, total_loss = -242.0 (0.68)\n",
            "epoch: 17115, dloss = 1.4, gloss = -2357.7, total_loss = -235.1 (0.66)\n",
            "epoch: 17116, dloss = -1.9, gloss = -2364.7, total_loss = -237.4 (0.68)\n",
            "epoch: 17117, dloss = 2.6, gloss = -2312.3, total_loss = -229.9 (0.66)\n",
            "epoch: 17118, dloss = 0.1, gloss = -2336.2, total_loss = -233.6 (0.86)\n",
            "epoch: 17119, dloss = -0.7, gloss = -2387.8, total_loss = -239.1 (0.86)\n",
            "epoch: 17120, dloss = 2.9, gloss = -2339.9, total_loss = -232.5 (0.68)\n",
            "epoch: 17121, dloss = 1.8, gloss = -2405.2, total_loss = -239.6 (0.69)\n",
            "epoch: 17122, dloss = -1.9, gloss = -2438.8, total_loss = -244.9 (0.69)\n",
            "epoch: 17123, dloss = 2.1, gloss = -2478.3, total_loss = -246.8 (0.69)\n",
            "epoch: 17124, dloss = -0.1, gloss = -2447.4, total_loss = -244.8 (0.69)\n",
            "epoch: 17125, dloss = 1.6, gloss = -2358.8, total_loss = -235.1 (0.70)\n",
            "epoch: 17126, dloss = 2.0, gloss = -2409.2, total_loss = -239.9 (0.66)\n",
            "epoch: 17127, dloss = -2.3, gloss = -2341.1, total_loss = -235.2 (0.70)\n",
            "epoch: 17128, dloss = 3.1, gloss = -2473.3, total_loss = -245.8 (0.69)\n",
            "epoch: 17129, dloss = 0.2, gloss = -2440.6, total_loss = -244.0 (0.68)\n",
            "epoch: 17130, dloss = 1.2, gloss = -2420.7, total_loss = -241.5 (0.69)\n",
            "epoch: 17131, dloss = 1.2, gloss = -2421.8, total_loss = -241.6 (0.68)\n",
            "epoch: 17132, dloss = -1.1, gloss = -2418.0, total_loss = -242.4 (0.67)\n",
            "epoch: 17133, dloss = 3.4, gloss = -2381.9, total_loss = -236.5 (0.69)\n",
            "epoch: 17134, dloss = 0.6, gloss = -2420.5, total_loss = -241.7 (0.98)\n",
            "epoch: 17135, dloss = 3.2, gloss = -2390.6, total_loss = -237.5 (0.72)\n",
            "epoch: 17136, dloss = -3.6, gloss = -2388.0, total_loss = -240.6 (0.66)\n",
            "epoch: 17137, dloss = -1.7, gloss = -2396.9, total_loss = -240.5 (0.70)\n",
            "epoch: 17138, dloss = 1.8, gloss = -2429.3, total_loss = -242.0 (0.69)\n",
            "epoch: 17139, dloss = -1.0, gloss = -2452.2, total_loss = -245.7 (0.67)\n",
            "epoch: 17140, dloss = 0.4, gloss = -2484.2, total_loss = -248.2 (0.73)\n",
            "epoch: 17141, dloss = 2.0, gloss = -2426.8, total_loss = -241.7 (0.69)\n",
            "epoch: 17142, dloss = -2.1, gloss = -2391.9, total_loss = -240.2 (0.70)\n",
            "epoch: 17143, dloss = -1.1, gloss = -2375.2, total_loss = -238.1 (0.70)\n",
            "epoch: 17144, dloss = -0.8, gloss = -2374.5, total_loss = -237.9 (0.69)\n",
            "epoch: 17145, dloss = 0.8, gloss = -2458.3, total_loss = -245.5 (0.68)\n",
            "epoch: 17146, dloss = -1.6, gloss = -2469.9, total_loss = -247.8 (0.67)\n",
            "epoch: 17147, dloss = -1.7, gloss = -2402.1, total_loss = -241.0 (0.71)\n",
            "epoch: 17148, dloss = 2.9, gloss = -2459.5, total_loss = -244.5 (0.70)\n",
            "epoch: 17149, dloss = 2.9, gloss = -2415.0, total_loss = -240.0 (0.86)\n",
            "epoch: 17150, dloss = 2.8, gloss = -2413.6, total_loss = -240.0 (0.90)\n",
            "epoch: 17151, dloss = 2.5, gloss = -2402.3, total_loss = -239.0 (0.68)\n",
            "epoch: 17152, dloss = 1.8, gloss = -2417.3, total_loss = -240.8 (0.68)\n",
            "epoch: 17153, dloss = 2.0, gloss = -2392.8, total_loss = -238.3 (0.68)\n",
            "epoch: 17154, dloss = 2.8, gloss = -2400.0, total_loss = -238.6 (0.70)\n",
            "epoch: 17155, dloss = 0.2, gloss = -2424.6, total_loss = -242.4 (0.71)\n",
            "epoch: 17156, dloss = 1.8, gloss = -2446.7, total_loss = -243.8 (0.68)\n",
            "epoch: 17157, dloss = 0.1, gloss = -2421.5, total_loss = -242.1 (0.68)\n",
            "epoch: 17158, dloss = 1.9, gloss = -2378.3, total_loss = -236.9 (0.72)\n",
            "epoch: 17159, dloss = -0.6, gloss = -2467.8, total_loss = -247.1 (0.70)\n",
            "epoch: 17160, dloss = 3.2, gloss = -2439.1, total_loss = -242.3 (0.68)\n",
            "epoch: 17161, dloss = 2.4, gloss = -2442.8, total_loss = -243.1 (0.69)\n",
            "epoch: 17162, dloss = 1.3, gloss = -2475.9, total_loss = -247.0 (0.70)\n",
            "epoch: 17163, dloss = 3.6, gloss = -2486.5, total_loss = -246.8 (0.67)\n",
            "epoch: 17164, dloss = 3.6, gloss = -2413.8, total_loss = -239.6 (0.69)\n",
            "epoch: 17165, dloss = 3.7, gloss = -2404.7, total_loss = -238.6 (1.07)\n",
            "epoch: 17166, dloss = 2.3, gloss = -2448.7, total_loss = -243.7 (0.73)\n",
            "epoch: 17167, dloss = -0.9, gloss = -2389.6, total_loss = -239.4 (0.70)\n",
            "epoch: 17168, dloss = 2.2, gloss = -2376.1, total_loss = -236.5 (0.67)\n",
            "epoch: 17169, dloss = 0.4, gloss = -2477.5, total_loss = -247.6 (0.72)\n",
            "epoch: 17170, dloss = 1.4, gloss = -2525.3, total_loss = -251.8 (0.69)\n",
            "epoch: 17171, dloss = -0.5, gloss = -2529.7, total_loss = -253.2 (0.67)\n",
            "epoch: 17172, dloss = 0.3, gloss = -2480.5, total_loss = -247.9 (0.68)\n",
            "epoch: 17173, dloss = 1.1, gloss = -2442.5, total_loss = -243.7 (0.67)\n",
            "epoch: 17174, dloss = 2.5, gloss = -2333.9, total_loss = -232.1 (0.69)\n",
            "epoch: 17175, dloss = 0.8, gloss = -2412.1, total_loss = -240.8 (0.70)\n",
            "epoch: 17176, dloss = -5.7, gloss = -2392.2, total_loss = -242.0 (0.68)\n",
            "epoch: 17177, dloss = 1.3, gloss = -2421.1, total_loss = -241.5 (0.68)\n",
            "epoch: 17178, dloss = -2.1, gloss = -2446.1, total_loss = -245.6 (0.71)\n",
            "epoch: 17179, dloss = 3.5, gloss = -2459.6, total_loss = -244.2 (0.67)\n",
            "epoch: 17180, dloss = 2.2, gloss = -2384.3, total_loss = -237.3 (0.87)\n",
            "epoch: 17181, dloss = 1.1, gloss = -2421.8, total_loss = -241.6 (0.86)\n",
            "epoch: 17182, dloss = 5.0, gloss = -2430.9, total_loss = -240.6 (0.69)\n",
            "epoch: 17183, dloss = -2.5, gloss = -2382.5, total_loss = -239.5 (0.66)\n",
            "epoch: 17184, dloss = 1.4, gloss = -2479.0, total_loss = -247.2 (0.68)\n",
            "epoch: 17185, dloss = -1.8, gloss = -2406.0, total_loss = -241.5 (0.68)\n",
            "epoch: 17186, dloss = -0.4, gloss = -2498.9, total_loss = -250.1 (0.67)\n",
            "epoch: 17187, dloss = 0.3, gloss = -2413.8, total_loss = -241.3 (0.68)\n",
            "epoch: 17188, dloss = 2.1, gloss = -2395.3, total_loss = -238.5 (0.68)\n",
            "epoch: 17189, dloss = 0.6, gloss = -2435.3, total_loss = -243.2 (0.67)\n",
            "epoch: 17190, dloss = -2.3, gloss = -2424.8, total_loss = -243.6 (0.67)\n",
            "epoch: 17191, dloss = -3.1, gloss = -2431.4, total_loss = -244.7 (0.70)\n",
            "epoch: 17192, dloss = 2.1, gloss = -2495.9, total_loss = -248.5 (0.67)\n",
            "epoch: 17193, dloss = -1.4, gloss = -2435.4, total_loss = -244.3 (0.69)\n",
            "epoch: 17194, dloss = -0.5, gloss = -2449.1, total_loss = -245.1 (0.69)\n",
            "epoch: 17195, dloss = 0.2, gloss = -2523.0, total_loss = -252.2 (0.66)\n",
            "epoch: 17196, dloss = 0.1, gloss = -2439.6, total_loss = -243.9 (0.95)\n",
            "epoch: 17197, dloss = 2.9, gloss = -2467.5, total_loss = -245.3 (0.76)\n",
            "epoch: 17198, dloss = -1.4, gloss = -2390.1, total_loss = -239.7 (0.67)\n",
            "epoch: 17199, dloss = -0.3, gloss = -2428.1, total_loss = -243.0 (0.68)\n",
            "epoch: 17200, dloss = -1.7, gloss = -2497.0, total_loss = -250.5 (0.69)\n",
            "epoch: 17201, dloss = 2.1, gloss = -2418.8, total_loss = -240.8 (0.69)\n",
            "epoch: 17202, dloss = 3.3, gloss = -2443.4, total_loss = -242.7 (0.68)\n",
            "epoch: 17203, dloss = 0.0, gloss = -2362.3, total_loss = -236.2 (0.71)\n",
            "epoch: 17204, dloss = 1.0, gloss = -2395.0, total_loss = -239.0 (0.69)\n",
            "epoch: 17205, dloss = -1.6, gloss = -2354.7, total_loss = -236.3 (0.69)\n",
            "epoch: 17206, dloss = 2.5, gloss = -2402.8, total_loss = -239.0 (0.71)\n",
            "epoch: 17207, dloss = 3.7, gloss = -2441.1, total_loss = -242.2 (0.69)\n",
            "epoch: 17208, dloss = 3.0, gloss = -2507.1, total_loss = -249.2 (0.69)\n",
            "epoch: 17209, dloss = 0.1, gloss = -2463.0, total_loss = -246.3 (0.72)\n",
            "epoch: 17210, dloss = -0.1, gloss = -2473.9, total_loss = -247.5 (0.73)\n",
            "epoch: 17211, dloss = 2.8, gloss = -2407.2, total_loss = -239.3 (0.82)\n",
            "epoch: 17212, dloss = 1.8, gloss = -2378.6, total_loss = -237.0 (0.94)\n",
            "epoch: 17213, dloss = 1.6, gloss = -2373.6, total_loss = -236.6 (0.69)\n",
            "epoch: 17214, dloss = 0.7, gloss = -2301.2, total_loss = -229.7 (0.70)\n",
            "epoch: 17215, dloss = 3.2, gloss = -2394.8, total_loss = -237.9 (0.68)\n",
            "epoch: 17216, dloss = 0.7, gloss = -2289.2, total_loss = -228.6 (0.69)\n",
            "epoch: 17217, dloss = 1.3, gloss = -2323.9, total_loss = -231.7 (0.68)\n",
            "epoch: 17218, dloss = 3.0, gloss = -2317.2, total_loss = -230.2 (0.71)\n",
            "epoch: 17219, dloss = -1.8, gloss = -2309.8, total_loss = -231.9 (0.68)\n",
            "epoch: 17220, dloss = 3.2, gloss = -2297.6, total_loss = -228.1 (0.70)\n",
            "epoch: 17221, dloss = 2.2, gloss = -2448.7, total_loss = -243.8 (0.69)\n",
            "epoch: 17222, dloss = 2.3, gloss = -2416.1, total_loss = -240.5 (0.67)\n",
            "epoch: 17223, dloss = -0.1, gloss = -2323.2, total_loss = -232.4 (0.68)\n",
            "epoch: 17224, dloss = -5.7, gloss = -2280.3, total_loss = -230.9 (0.69)\n",
            "epoch: 17225, dloss = -0.0, gloss = -2341.2, total_loss = -234.1 (0.67)\n",
            "epoch: 17226, dloss = 2.5, gloss = -2380.9, total_loss = -236.8 (0.69)\n",
            "epoch: 17227, dloss = 1.6, gloss = -2402.0, total_loss = -239.4 (0.98)\n",
            "epoch: 17228, dloss = 3.9, gloss = -2346.2, total_loss = -232.7 (0.78)\n",
            "epoch: 17229, dloss = 4.1, gloss = -2479.1, total_loss = -245.9 (0.69)\n",
            "epoch: 17230, dloss = 2.4, gloss = -2457.5, total_loss = -244.6 (0.68)\n",
            "epoch: 17231, dloss = 2.4, gloss = -2352.0, total_loss = -234.0 (0.70)\n",
            "epoch: 17232, dloss = 1.7, gloss = -2404.5, total_loss = -239.6 (0.67)\n",
            "epoch: 17233, dloss = -0.6, gloss = -2432.2, total_loss = -243.5 (0.68)\n",
            "epoch: 17234, dloss = 2.2, gloss = -2386.5, total_loss = -237.6 (0.70)\n",
            "epoch: 17235, dloss = 0.8, gloss = -2388.0, total_loss = -238.4 (0.68)\n",
            "epoch: 17236, dloss = -0.6, gloss = -2468.6, total_loss = -247.2 (0.66)\n",
            "epoch: 17237, dloss = 6.1, gloss = -2412.0, total_loss = -238.2 (0.69)\n",
            "epoch: 17238, dloss = 1.3, gloss = -2406.6, total_loss = -240.0 (0.68)\n",
            "epoch: 17239, dloss = -0.6, gloss = -2409.0, total_loss = -241.2 (0.67)\n",
            "epoch: 17240, dloss = 1.6, gloss = -2448.3, total_loss = -244.0 (0.71)\n",
            "epoch: 17241, dloss = 1.7, gloss = -2472.4, total_loss = -246.4 (0.70)\n",
            "epoch: 17242, dloss = -0.1, gloss = -2501.5, total_loss = -250.2 (0.85)\n",
            "epoch: 17243, dloss = 1.2, gloss = -2522.5, total_loss = -251.7 (0.93)\n",
            "epoch: 17244, dloss = 1.4, gloss = -2445.5, total_loss = -243.9 (0.71)\n",
            "epoch: 17245, dloss = -1.5, gloss = -2437.8, total_loss = -244.5 (0.73)\n",
            "epoch: 17246, dloss = 0.6, gloss = -2409.2, total_loss = -240.6 (0.69)\n",
            "epoch: 17247, dloss = 0.4, gloss = -2453.6, total_loss = -245.2 (0.72)\n",
            "epoch: 17248, dloss = -0.6, gloss = -2500.3, total_loss = -250.3 (0.69)\n",
            "epoch: 17249, dloss = 3.0, gloss = -2558.3, total_loss = -254.3 (0.66)\n",
            "epoch: 17250, dloss = 0.3, gloss = -2604.9, total_loss = -260.4 (0.69)\n",
            "epoch: 17251, dloss = 2.3, gloss = -2490.0, total_loss = -247.9 (0.72)\n",
            "epoch: 17252, dloss = 1.8, gloss = -2449.5, total_loss = -244.0 (0.69)\n",
            "epoch: 17253, dloss = 0.3, gloss = -2441.8, total_loss = -244.1 (0.69)\n",
            "epoch: 17254, dloss = 1.5, gloss = -2376.1, total_loss = -236.8 (0.71)\n",
            "epoch: 17255, dloss = -0.3, gloss = -2417.3, total_loss = -241.9 (0.67)\n",
            "epoch: 17256, dloss = 2.0, gloss = -2397.7, total_loss = -238.8 (0.69)\n",
            "epoch: 17257, dloss = 1.5, gloss = -2441.3, total_loss = -243.4 (0.70)\n",
            "epoch: 17258, dloss = 1.5, gloss = -2439.0, total_loss = -243.2 (1.07)\n",
            "epoch: 17259, dloss = 1.9, gloss = -2472.7, total_loss = -246.3 (0.71)\n",
            "epoch: 17260, dloss = -0.1, gloss = -2403.1, total_loss = -240.4 (0.69)\n",
            "epoch: 17261, dloss = -0.5, gloss = -2449.9, total_loss = -245.3 (0.69)\n",
            "epoch: 17262, dloss = 2.3, gloss = -2476.2, total_loss = -246.5 (0.82)\n",
            "epoch: 17263, dloss = 1.6, gloss = -2389.6, total_loss = -238.2 (1.03)\n",
            "epoch: 17264, dloss = -1.6, gloss = -2369.6, total_loss = -237.7 (0.68)\n",
            "epoch: 17265, dloss = 2.7, gloss = -2448.8, total_loss = -243.5 (0.69)\n",
            "epoch: 17266, dloss = 2.5, gloss = -2544.1, total_loss = -253.2 (0.66)\n",
            "epoch: 17267, dloss = 6.2, gloss = -2580.2, total_loss = -254.9 (0.67)\n",
            "epoch: 17268, dloss = 1.2, gloss = -2599.3, total_loss = -259.3 (0.70)\n",
            "epoch: 17269, dloss = 3.4, gloss = -2599.4, total_loss = -258.2 (0.70)\n",
            "epoch: 17270, dloss = 1.5, gloss = -2546.6, total_loss = -253.9 (0.68)\n",
            "epoch: 17271, dloss = 1.1, gloss = -2552.8, total_loss = -254.7 (0.68)\n",
            "epoch: 17272, dloss = 4.3, gloss = -2640.6, total_loss = -261.9 (0.73)\n",
            "epoch: 17273, dloss = 2.9, gloss = -2521.0, total_loss = -250.7 (0.97)\n",
            "epoch: 17274, dloss = 2.3, gloss = -2488.3, total_loss = -247.7 (0.68)\n",
            "epoch: 17275, dloss = 5.1, gloss = -2576.4, total_loss = -255.1 (0.70)\n",
            "epoch: 17276, dloss = 2.6, gloss = -2470.2, total_loss = -245.7 (0.66)\n",
            "epoch: 17277, dloss = -1.5, gloss = -2532.7, total_loss = -254.0 (0.71)\n",
            "epoch: 17278, dloss = 2.5, gloss = -2474.5, total_loss = -246.2 (0.68)\n",
            "epoch: 17279, dloss = 1.7, gloss = -2497.3, total_loss = -248.9 (0.69)\n",
            "epoch: 17280, dloss = 2.6, gloss = -2549.0, total_loss = -253.6 (0.68)\n",
            "epoch: 17281, dloss = 2.4, gloss = -2464.6, total_loss = -245.2 (0.70)\n",
            "epoch: 17282, dloss = 1.6, gloss = -2437.2, total_loss = -242.9 (0.69)\n",
            "epoch: 17283, dloss = 0.9, gloss = -2469.0, total_loss = -246.4 (0.70)\n",
            "epoch: 17284, dloss = -0.1, gloss = -2471.6, total_loss = -247.2 (0.70)\n",
            "epoch: 17285, dloss = 1.7, gloss = -2467.6, total_loss = -245.9 (0.70)\n",
            "epoch: 17286, dloss = 2.3, gloss = -2445.2, total_loss = -243.4 (0.68)\n",
            "epoch: 17287, dloss = 2.7, gloss = -2428.2, total_loss = -241.5 (0.67)\n",
            "epoch: 17288, dloss = 0.4, gloss = -2418.1, total_loss = -241.6 (0.90)\n",
            "epoch: 17289, dloss = 4.3, gloss = -2495.7, total_loss = -247.4 (0.87)\n",
            "epoch: 17290, dloss = 1.4, gloss = -2485.8, total_loss = -247.9 (0.68)\n",
            "epoch: 17291, dloss = 4.5, gloss = -2436.1, total_loss = -241.4 (0.71)\n",
            "epoch: 17292, dloss = 4.7, gloss = -2386.2, total_loss = -236.3 (0.68)\n",
            "epoch: 17293, dloss = 3.6, gloss = -2466.7, total_loss = -244.9 (0.70)\n",
            "epoch: 17294, dloss = 2.0, gloss = -2457.5, total_loss = -244.8 (0.68)\n",
            "epoch: 17295, dloss = 0.2, gloss = -2473.5, total_loss = -247.2 (0.70)\n",
            "epoch: 17296, dloss = 4.1, gloss = -2567.1, total_loss = -254.7 (0.69)\n",
            "epoch: 17297, dloss = 2.7, gloss = -2545.3, total_loss = -253.2 (0.67)\n",
            "epoch: 17298, dloss = 1.3, gloss = -2483.3, total_loss = -247.7 (0.70)\n",
            "epoch: 17299, dloss = 0.4, gloss = -2431.5, total_loss = -242.9 (0.70)\n",
            "epoch: 17300, dloss = 1.1, gloss = -2460.6, total_loss = -245.5 (0.68)\n",
            "epoch: 17301, dloss = 2.0, gloss = -2477.4, total_loss = -246.7 (0.67)\n",
            "epoch: 17302, dloss = -1.5, gloss = -2481.5, total_loss = -248.9 (0.70)\n",
            "epoch: 17303, dloss = 1.5, gloss = -2493.1, total_loss = -248.6 (0.73)\n",
            "epoch: 17304, dloss = 2.6, gloss = -2465.3, total_loss = -245.2 (1.00)\n",
            "epoch: 17305, dloss = -0.3, gloss = -2501.1, total_loss = -250.3 (0.69)\n",
            "epoch: 17306, dloss = 0.9, gloss = -2503.7, total_loss = -249.9 (0.68)\n",
            "epoch: 17307, dloss = 1.7, gloss = -2414.7, total_loss = -240.6 (0.67)\n",
            "epoch: 17308, dloss = 0.3, gloss = -2450.8, total_loss = -244.9 (0.69)\n",
            "epoch: 17309, dloss = 0.8, gloss = -2557.4, total_loss = -255.4 (0.70)\n",
            "epoch: 17310, dloss = 3.5, gloss = -2538.5, total_loss = -252.1 (0.68)\n",
            "epoch: 17311, dloss = 1.3, gloss = -2485.5, total_loss = -247.9 (0.68)\n",
            "epoch: 17312, dloss = 2.3, gloss = -2451.0, total_loss = -244.0 (0.70)\n",
            "epoch: 17313, dloss = 2.4, gloss = -2491.6, total_loss = -248.0 (0.98)\n",
            "epoch: 17314, dloss = 5.2, gloss = -2489.1, total_loss = -246.3 (0.79)\n",
            "epoch: 17315, dloss = 1.7, gloss = -2425.4, total_loss = -241.7 (0.69)\n",
            "epoch: 17316, dloss = 1.7, gloss = -2461.7, total_loss = -245.3 (0.70)\n",
            "epoch: 17317, dloss = 3.0, gloss = -2441.1, total_loss = -242.6 (0.74)\n",
            "epoch: 17318, dloss = 2.5, gloss = -2435.7, total_loss = -242.3 (1.14)\n",
            "epoch: 17319, dloss = 1.3, gloss = -2446.5, total_loss = -244.0 (0.94)\n",
            "epoch: 17320, dloss = 3.1, gloss = -2476.1, total_loss = -246.1 (0.69)\n",
            "epoch: 17321, dloss = 3.2, gloss = -2410.0, total_loss = -239.4 (0.69)\n",
            "epoch: 17322, dloss = 1.1, gloss = -2450.8, total_loss = -244.5 (0.70)\n",
            "epoch: 17323, dloss = 1.8, gloss = -2440.2, total_loss = -243.1 (0.69)\n",
            "epoch: 17324, dloss = 2.7, gloss = -2530.2, total_loss = -251.7 (0.69)\n",
            "epoch: 17325, dloss = 1.5, gloss = -2498.9, total_loss = -249.2 (0.68)\n",
            "epoch: 17326, dloss = 5.6, gloss = -2517.9, total_loss = -249.0 (0.66)\n",
            "epoch: 17327, dloss = 1.4, gloss = -2536.3, total_loss = -252.9 (0.68)\n",
            "epoch: 17328, dloss = 0.8, gloss = -2565.7, total_loss = -256.2 (0.70)\n",
            "epoch: 17329, dloss = 1.9, gloss = -2519.2, total_loss = -250.9 (0.68)\n",
            "epoch: 17330, dloss = -1.0, gloss = -2552.2, total_loss = -255.7 (0.69)\n",
            "epoch: 17331, dloss = 1.4, gloss = -2556.0, total_loss = -254.9 (0.69)\n",
            "epoch: 17332, dloss = -1.1, gloss = -2553.0, total_loss = -255.9 (0.68)\n",
            "epoch: 17333, dloss = -0.8, gloss = -2514.9, total_loss = -251.9 (0.68)\n",
            "epoch: 17334, dloss = 2.0, gloss = -2513.5, total_loss = -250.3 (1.01)\n",
            "epoch: 17335, dloss = -1.3, gloss = -2487.2, total_loss = -249.4 (0.77)\n",
            "epoch: 17336, dloss = 0.2, gloss = -2508.6, total_loss = -250.8 (0.68)\n",
            "epoch: 17337, dloss = 2.6, gloss = -2500.0, total_loss = -248.7 (0.68)\n",
            "epoch: 17338, dloss = 1.5, gloss = -2431.2, total_loss = -242.4 (0.70)\n",
            "epoch: 17339, dloss = 1.2, gloss = -2436.4, total_loss = -243.1 (0.69)\n",
            "epoch: 17340, dloss = 0.5, gloss = -2450.6, total_loss = -244.8 (0.68)\n",
            "epoch: 17341, dloss = 2.5, gloss = -2508.9, total_loss = -249.6 (0.71)\n",
            "epoch: 17342, dloss = 2.0, gloss = -2484.2, total_loss = -247.4 (0.68)\n",
            "epoch: 17343, dloss = 2.6, gloss = -2562.4, total_loss = -254.9 (0.68)\n",
            "epoch: 17344, dloss = 2.9, gloss = -2553.3, total_loss = -253.9 (0.67)\n",
            "epoch: 17345, dloss = 1.9, gloss = -2552.1, total_loss = -254.3 (0.67)\n",
            "epoch: 17346, dloss = 0.8, gloss = -2511.1, total_loss = -250.7 (0.69)\n",
            "epoch: 17347, dloss = 2.5, gloss = -2457.3, total_loss = -244.5 (0.70)\n",
            "epoch: 17348, dloss = -1.1, gloss = -2422.8, total_loss = -242.8 (0.70)\n",
            "epoch: 17349, dloss = 2.3, gloss = -2443.7, total_loss = -243.2 (0.83)\n",
            "epoch: 17350, dloss = 2.8, gloss = -2585.7, total_loss = -257.2 (0.94)\n",
            "epoch: 17351, dloss = 3.4, gloss = -2568.9, total_loss = -255.2 (0.69)\n",
            "epoch: 17352, dloss = 3.0, gloss = -2534.1, total_loss = -251.9 (0.68)\n",
            "epoch: 17353, dloss = 1.6, gloss = -2504.1, total_loss = -249.6 (0.68)\n",
            "epoch: 17354, dloss = 3.9, gloss = -2574.5, total_loss = -255.5 (0.69)\n",
            "epoch: 17355, dloss = -0.2, gloss = -2543.7, total_loss = -254.5 (0.69)\n",
            "epoch: 17356, dloss = -1.6, gloss = -2548.9, total_loss = -255.7 (0.71)\n",
            "epoch: 17357, dloss = 3.2, gloss = -2487.5, total_loss = -247.2 (0.68)\n",
            "epoch: 17358, dloss = 0.5, gloss = -2475.6, total_loss = -247.3 (0.69)\n",
            "epoch: 17359, dloss = 3.6, gloss = -2525.4, total_loss = -250.8 (0.67)\n",
            "epoch: 17360, dloss = 1.7, gloss = -2544.7, total_loss = -253.6 (0.70)\n",
            "epoch: 17361, dloss = -1.4, gloss = -2514.3, total_loss = -252.1 (0.66)\n",
            "epoch: 17362, dloss = 0.9, gloss = -2577.2, total_loss = -257.3 (0.68)\n",
            "epoch: 17363, dloss = 1.5, gloss = -2578.5, total_loss = -257.1 (0.68)\n",
            "epoch: 17364, dloss = 1.7, gloss = -2532.6, total_loss = -252.4 (0.69)\n",
            "epoch: 17365, dloss = -1.2, gloss = -2494.7, total_loss = -250.0 (0.93)\n",
            "epoch: 17366, dloss = -0.5, gloss = -2537.2, total_loss = -254.0 (0.79)\n",
            "epoch: 17367, dloss = 2.4, gloss = -2541.5, total_loss = -253.0 (0.66)\n",
            "epoch: 17368, dloss = 5.1, gloss = -2499.3, total_loss = -247.4 (0.68)\n",
            "epoch: 17369, dloss = -0.1, gloss = -2427.1, total_loss = -242.8 (0.67)\n",
            "epoch: 17370, dloss = 4.2, gloss = -2465.7, total_loss = -244.4 (0.69)\n",
            "epoch: 17371, dloss = 1.2, gloss = -2399.0, total_loss = -239.3 (0.69)\n",
            "epoch: 17372, dloss = 1.1, gloss = -2402.7, total_loss = -239.7 (0.69)\n",
            "epoch: 17373, dloss = 0.5, gloss = -2435.1, total_loss = -243.2 (0.67)\n",
            "epoch: 17374, dloss = 3.1, gloss = -2453.4, total_loss = -243.8 (0.72)\n",
            "epoch: 17375, dloss = -0.5, gloss = -2455.7, total_loss = -245.8 (0.70)\n",
            "epoch: 17376, dloss = 0.4, gloss = -2448.0, total_loss = -244.6 (0.70)\n",
            "epoch: 17377, dloss = -2.2, gloss = -2474.8, total_loss = -248.6 (0.68)\n",
            "epoch: 17378, dloss = -3.8, gloss = -2573.2, total_loss = -259.2 (0.69)\n",
            "epoch: 17379, dloss = 1.3, gloss = -2528.1, total_loss = -252.2 (0.68)\n",
            "epoch: 17380, dloss = 1.8, gloss = -2441.7, total_loss = -243.3 (0.73)\n",
            "epoch: 17381, dloss = 4.6, gloss = -2471.6, total_loss = -244.8 (0.99)\n",
            "epoch: 17382, dloss = 2.8, gloss = -2509.3, total_loss = -249.5 (0.69)\n",
            "epoch: 17383, dloss = 2.6, gloss = -2559.4, total_loss = -254.6 (0.68)\n",
            "epoch: 17384, dloss = 2.8, gloss = -2509.2, total_loss = -249.5 (0.67)\n",
            "epoch: 17385, dloss = 3.2, gloss = -2513.8, total_loss = -249.8 (0.66)\n",
            "epoch: 17386, dloss = 3.3, gloss = -2503.8, total_loss = -248.8 (0.69)\n",
            "epoch: 17387, dloss = 2.2, gloss = -2525.7, total_loss = -251.5 (0.68)\n",
            "epoch: 17388, dloss = 3.0, gloss = -2511.6, total_loss = -249.7 (0.67)\n",
            "epoch: 17389, dloss = 3.2, gloss = -2452.7, total_loss = -243.7 (0.68)\n",
            "epoch: 17390, dloss = 1.2, gloss = -2502.7, total_loss = -249.7 (0.70)\n",
            "epoch: 17391, dloss = 3.1, gloss = -2526.8, total_loss = -251.1 (0.68)\n",
            "epoch: 17392, dloss = 2.4, gloss = -2612.3, total_loss = -260.0 (0.67)\n",
            "epoch: 17393, dloss = 4.6, gloss = -2571.4, total_loss = -254.8 (0.68)\n",
            "epoch: 17394, dloss = 3.8, gloss = -2488.1, total_loss = -246.9 (0.68)\n",
            "epoch: 17395, dloss = 2.2, gloss = -2539.1, total_loss = -252.8 (0.68)\n",
            "epoch: 17396, dloss = 0.1, gloss = -2547.6, total_loss = -254.7 (0.82)\n",
            "epoch: 17397, dloss = 0.7, gloss = -2513.1, total_loss = -251.0 (0.91)\n",
            "epoch: 17398, dloss = 3.0, gloss = -2584.2, total_loss = -256.9 (0.68)\n",
            "epoch: 17399, dloss = 2.1, gloss = -2499.2, total_loss = -248.9 (0.70)\n",
            "epoch: 17400, dloss = 3.5, gloss = -2443.1, total_loss = -242.6 (0.69)\n",
            "epoch: 17401, dloss = -1.0, gloss = -2509.3, total_loss = -251.4 (0.66)\n",
            "epoch: 17402, dloss = 0.3, gloss = -2509.0, total_loss = -250.7 (0.68)\n",
            "epoch: 17403, dloss = 0.1, gloss = -2528.9, total_loss = -252.8 (0.69)\n",
            "epoch: 17404, dloss = 2.1, gloss = -2585.8, total_loss = -257.5 (0.65)\n",
            "epoch: 17405, dloss = -0.0, gloss = -2512.5, total_loss = -251.3 (0.70)\n",
            "epoch: 17406, dloss = 2.4, gloss = -2499.8, total_loss = -248.8 (0.69)\n",
            "epoch: 17407, dloss = 1.8, gloss = -2560.8, total_loss = -255.2 (0.68)\n",
            "epoch: 17408, dloss = -1.0, gloss = -2577.4, total_loss = -258.2 (0.69)\n",
            "epoch: 17409, dloss = 3.0, gloss = -2573.6, total_loss = -255.8 (0.71)\n",
            "epoch: 17410, dloss = 0.9, gloss = -2537.4, total_loss = -253.3 (0.67)\n",
            "epoch: 17411, dloss = 3.7, gloss = -2518.1, total_loss = -249.9 (0.70)\n",
            "epoch: 17412, dloss = 2.8, gloss = -2528.9, total_loss = -251.5 (1.01)\n",
            "epoch: 17413, dloss = 2.5, gloss = -2488.5, total_loss = -247.6 (0.75)\n",
            "epoch: 17414, dloss = 3.0, gloss = -2444.7, total_loss = -243.0 (0.66)\n",
            "epoch: 17415, dloss = 4.2, gloss = -2442.4, total_loss = -242.1 (0.69)\n",
            "epoch: 17416, dloss = 2.7, gloss = -2426.4, total_loss = -241.3 (0.68)\n",
            "epoch: 17417, dloss = 1.6, gloss = -2519.1, total_loss = -251.1 (0.68)\n",
            "epoch: 17418, dloss = 3.0, gloss = -2564.7, total_loss = -255.0 (0.68)\n",
            "epoch: 17419, dloss = 1.1, gloss = -2506.6, total_loss = -250.1 (0.68)\n",
            "epoch: 17420, dloss = 1.3, gloss = -2470.7, total_loss = -246.4 (0.66)\n",
            "epoch: 17421, dloss = 5.2, gloss = -2503.0, total_loss = -247.7 (0.68)\n",
            "epoch: 17422, dloss = 0.1, gloss = -2491.8, total_loss = -249.1 (0.70)\n",
            "epoch: 17423, dloss = -0.1, gloss = -2503.4, total_loss = -250.4 (0.68)\n",
            "epoch: 17424, dloss = 2.5, gloss = -2462.7, total_loss = -245.0 (0.70)\n",
            "epoch: 17425, dloss = -0.5, gloss = -2546.3, total_loss = -254.9 (0.69)\n",
            "epoch: 17426, dloss = 1.0, gloss = -2480.2, total_loss = -247.5 (0.68)\n",
            "epoch: 17427, dloss = 2.4, gloss = -2476.8, total_loss = -246.5 (0.74)\n",
            "epoch: 17428, dloss = 2.2, gloss = -2463.2, total_loss = -245.2 (0.98)\n",
            "epoch: 17429, dloss = 4.8, gloss = -2481.8, total_loss = -245.8 (0.68)\n",
            "epoch: 17430, dloss = 2.2, gloss = -2508.3, total_loss = -249.7 (0.67)\n",
            "epoch: 17431, dloss = 2.8, gloss = -2515.1, total_loss = -250.1 (0.67)\n",
            "epoch: 17432, dloss = 0.9, gloss = -2597.4, total_loss = -259.3 (0.68)\n",
            "epoch: 17433, dloss = 4.0, gloss = -2510.1, total_loss = -249.0 (0.68)\n",
            "epoch: 17434, dloss = 2.8, gloss = -2509.8, total_loss = -249.6 (0.69)\n",
            "epoch: 17435, dloss = -1.0, gloss = -2506.7, total_loss = -251.2 (0.68)\n",
            "epoch: 17436, dloss = 1.8, gloss = -2521.7, total_loss = -251.3 (0.67)\n",
            "epoch: 17437, dloss = 4.5, gloss = -2544.9, total_loss = -252.2 (0.67)\n",
            "epoch: 17438, dloss = 4.1, gloss = -2527.3, total_loss = -250.7 (0.67)\n",
            "epoch: 17439, dloss = 1.4, gloss = -2526.8, total_loss = -252.0 (0.66)\n",
            "epoch: 17440, dloss = 0.4, gloss = -2590.4, total_loss = -258.9 (0.67)\n",
            "epoch: 17441, dloss = 6.0, gloss = -2581.2, total_loss = -255.1 (0.68)\n",
            "epoch: 17442, dloss = 2.2, gloss = -2553.4, total_loss = -254.3 (0.68)\n",
            "epoch: 17443, dloss = 2.1, gloss = -2545.5, total_loss = -253.5 (0.83)\n",
            "epoch: 17444, dloss = 3.5, gloss = -2555.6, total_loss = -253.8 (0.88)\n",
            "epoch: 17445, dloss = 1.4, gloss = -2506.8, total_loss = -250.0 (0.67)\n",
            "epoch: 17446, dloss = 1.7, gloss = -2542.9, total_loss = -253.5 (0.67)\n",
            "epoch: 17447, dloss = 0.0, gloss = -2489.6, total_loss = -248.9 (0.66)\n",
            "epoch: 17448, dloss = -3.0, gloss = -2521.7, total_loss = -253.6 (0.68)\n",
            "epoch: 17449, dloss = 5.0, gloss = -2512.4, total_loss = -248.7 (0.65)\n",
            "epoch: 17450, dloss = 4.6, gloss = -2608.1, total_loss = -258.5 (0.67)\n",
            "epoch: 17451, dloss = 2.8, gloss = -2600.1, total_loss = -258.6 (0.68)\n",
            "epoch: 17452, dloss = 3.2, gloss = -2611.1, total_loss = -259.5 (0.67)\n",
            "epoch: 17453, dloss = 3.6, gloss = -2646.0, total_loss = -262.8 (0.68)\n",
            "epoch: 17454, dloss = 4.1, gloss = -2683.1, total_loss = -266.2 (0.69)\n",
            "epoch: 17455, dloss = 3.4, gloss = -2647.1, total_loss = -263.0 (0.69)\n",
            "epoch: 17456, dloss = 1.8, gloss = -2610.3, total_loss = -260.1 (0.68)\n",
            "epoch: 17457, dloss = 2.4, gloss = -2644.4, total_loss = -263.2 (0.66)\n",
            "epoch: 17458, dloss = 4.3, gloss = -2658.8, total_loss = -263.7 (0.68)\n",
            "epoch: 17459, dloss = 0.4, gloss = -2656.1, total_loss = -265.4 (0.88)\n",
            "epoch: 17460, dloss = 2.8, gloss = -2669.5, total_loss = -265.6 (0.86)\n",
            "epoch: 17461, dloss = 2.8, gloss = -2618.2, total_loss = -260.4 (0.69)\n",
            "epoch: 17462, dloss = 3.6, gloss = -2565.2, total_loss = -254.7 (0.69)\n",
            "epoch: 17463, dloss = 4.8, gloss = -2521.6, total_loss = -249.8 (0.68)\n",
            "epoch: 17464, dloss = 2.7, gloss = -2549.1, total_loss = -253.6 (0.67)\n",
            "epoch: 17465, dloss = 3.8, gloss = -2580.0, total_loss = -256.1 (0.68)\n",
            "epoch: 17466, dloss = 2.7, gloss = -2623.3, total_loss = -261.0 (0.68)\n",
            "epoch: 17467, dloss = 1.7, gloss = -2565.9, total_loss = -255.8 (0.67)\n",
            "epoch: 17468, dloss = 4.5, gloss = -2640.8, total_loss = -261.8 (0.67)\n",
            "epoch: 17469, dloss = 3.0, gloss = -2591.5, total_loss = -257.6 (0.69)\n",
            "epoch: 17470, dloss = 1.0, gloss = -2571.1, total_loss = -256.6 (0.66)\n",
            "epoch: 17471, dloss = 4.7, gloss = -2556.9, total_loss = -253.3 (0.67)\n",
            "epoch: 17472, dloss = 4.4, gloss = -2563.4, total_loss = -254.1 (0.68)\n",
            "epoch: 17473, dloss = 1.1, gloss = -2610.7, total_loss = -260.5 (0.68)\n",
            "epoch: 17474, dloss = 3.0, gloss = -2509.5, total_loss = -249.5 (0.67)\n",
            "epoch: 17475, dloss = 4.0, gloss = -2503.6, total_loss = -248.4 (0.96)\n",
            "epoch: 17476, dloss = 3.1, gloss = -2542.6, total_loss = -252.7 (0.77)\n",
            "epoch: 17477, dloss = 0.2, gloss = -2559.6, total_loss = -255.9 (0.68)\n",
            "epoch: 17478, dloss = 0.1, gloss = -2557.7, total_loss = -255.7 (0.67)\n",
            "epoch: 17479, dloss = 3.6, gloss = -2493.9, total_loss = -247.6 (0.68)\n",
            "epoch: 17480, dloss = 2.7, gloss = -2502.8, total_loss = -249.0 (0.69)\n",
            "epoch: 17481, dloss = -0.9, gloss = -2521.6, total_loss = -252.6 (0.71)\n",
            "epoch: 17482, dloss = 1.8, gloss = -2513.4, total_loss = -250.4 (0.66)\n",
            "epoch: 17483, dloss = 2.3, gloss = -2460.5, total_loss = -244.9 (0.68)\n",
            "epoch: 17484, dloss = 1.8, gloss = -2533.9, total_loss = -252.5 (0.67)\n",
            "epoch: 17485, dloss = 0.4, gloss = -2541.5, total_loss = -253.9 (0.69)\n",
            "epoch: 17486, dloss = 1.4, gloss = -2538.3, total_loss = -253.1 (0.69)\n",
            "epoch: 17487, dloss = 4.5, gloss = -2564.0, total_loss = -254.2 (0.67)\n",
            "epoch: 17488, dloss = 6.8, gloss = -2528.1, total_loss = -249.4 (0.69)\n",
            "epoch: 17489, dloss = 3.6, gloss = -2576.0, total_loss = -255.8 (0.68)\n",
            "epoch: 17490, dloss = 1.6, gloss = -2528.5, total_loss = -252.1 (0.75)\n",
            "epoch: 17491, dloss = 1.2, gloss = -2597.7, total_loss = -259.2 (0.96)\n",
            "epoch: 17492, dloss = 2.9, gloss = -2602.4, total_loss = -258.8 (0.69)\n",
            "epoch: 17493, dloss = 2.4, gloss = -2560.8, total_loss = -254.9 (0.69)\n",
            "epoch: 17494, dloss = 0.0, gloss = -2561.2, total_loss = -256.1 (0.69)\n",
            "epoch: 17495, dloss = -3.3, gloss = -2516.1, total_loss = -253.3 (0.68)\n",
            "epoch: 17496, dloss = -3.4, gloss = -2504.7, total_loss = -252.2 (0.69)\n",
            "epoch: 17497, dloss = 5.2, gloss = -2493.4, total_loss = -246.7 (0.68)\n",
            "epoch: 17498, dloss = 2.8, gloss = -2593.5, total_loss = -257.9 (0.72)\n",
            "epoch: 17499, dloss = 0.9, gloss = -2525.1, total_loss = -252.1 (0.70)\n",
            "epoch: 17500, dloss = 1.1, gloss = -2600.1, total_loss = -259.4 (0.70)\n",
            "epoch: 17501, dloss = 2.9, gloss = -2588.8, total_loss = -257.4 (0.69)\n",
            "epoch: 17502, dloss = -4.8, gloss = -2571.5, total_loss = -259.6 (0.71)\n",
            "epoch: 17503, dloss = 2.1, gloss = -2568.7, total_loss = -255.8 (0.70)\n",
            "epoch: 17504, dloss = 3.8, gloss = -2527.9, total_loss = -250.9 (0.67)\n",
            "epoch: 17505, dloss = 3.5, gloss = -2603.0, total_loss = -258.6 (0.72)\n",
            "epoch: 17506, dloss = 5.7, gloss = -2637.9, total_loss = -261.0 (0.99)\n",
            "epoch: 17507, dloss = 4.3, gloss = -2610.6, total_loss = -258.9 (0.77)\n",
            "epoch: 17508, dloss = 2.4, gloss = -2511.5, total_loss = -249.9 (0.68)\n",
            "epoch: 17509, dloss = 3.1, gloss = -2573.4, total_loss = -255.8 (0.69)\n",
            "epoch: 17510, dloss = 2.2, gloss = -2541.7, total_loss = -253.1 (0.69)\n",
            "epoch: 17511, dloss = 1.8, gloss = -2607.4, total_loss = -259.8 (0.70)\n",
            "epoch: 17512, dloss = 3.6, gloss = -2574.1, total_loss = -255.6 (0.69)\n",
            "epoch: 17513, dloss = 5.3, gloss = -2545.0, total_loss = -251.9 (0.69)\n",
            "epoch: 17514, dloss = -0.0, gloss = -2482.5, total_loss = -248.3 (0.69)\n",
            "epoch: 17515, dloss = 4.1, gloss = -2548.2, total_loss = -252.8 (0.73)\n",
            "epoch: 17516, dloss = 4.1, gloss = -2552.3, total_loss = -253.2 (0.70)\n",
            "epoch: 17517, dloss = -1.0, gloss = -2476.1, total_loss = -248.1 (0.66)\n",
            "epoch: 17518, dloss = 5.0, gloss = -2512.8, total_loss = -248.8 (0.69)\n",
            "epoch: 17519, dloss = 2.5, gloss = -2453.7, total_loss = -244.1 (0.67)\n",
            "epoch: 17520, dloss = 3.0, gloss = -2461.2, total_loss = -244.6 (0.66)\n",
            "epoch: 17521, dloss = 5.4, gloss = -2508.0, total_loss = -248.1 (0.80)\n",
            "epoch: 17522, dloss = 3.2, gloss = -2480.7, total_loss = -246.5 (0.92)\n",
            "epoch: 17523, dloss = -3.9, gloss = -2490.4, total_loss = -251.0 (0.68)\n",
            "epoch: 17524, dloss = 4.0, gloss = -2539.6, total_loss = -251.9 (0.70)\n",
            "epoch: 17525, dloss = 2.9, gloss = -2530.8, total_loss = -251.6 (0.69)\n",
            "epoch: 17526, dloss = 2.3, gloss = -2507.2, total_loss = -249.6 (0.69)\n",
            "epoch: 17527, dloss = 4.1, gloss = -2562.8, total_loss = -254.2 (0.69)\n",
            "epoch: 17528, dloss = 2.2, gloss = -2518.2, total_loss = -250.7 (0.66)\n",
            "epoch: 17529, dloss = -0.8, gloss = -2516.8, total_loss = -252.1 (0.69)\n",
            "epoch: 17530, dloss = 2.2, gloss = -2482.5, total_loss = -247.2 (0.66)\n",
            "epoch: 17531, dloss = 3.2, gloss = -2498.3, total_loss = -248.2 (0.69)\n",
            "epoch: 17532, dloss = 2.0, gloss = -2473.9, total_loss = -246.4 (0.68)\n",
            "epoch: 17533, dloss = -0.4, gloss = -2521.0, total_loss = -252.3 (0.68)\n",
            "epoch: 17534, dloss = 3.4, gloss = -2524.6, total_loss = -250.8 (0.70)\n",
            "epoch: 17535, dloss = 3.9, gloss = -2526.9, total_loss = -250.8 (0.69)\n",
            "epoch: 17536, dloss = 0.4, gloss = -2559.2, total_loss = -255.7 (0.72)\n",
            "epoch: 17537, dloss = 3.8, gloss = -2552.3, total_loss = -253.3 (0.97)\n",
            "epoch: 17538, dloss = 4.0, gloss = -2561.0, total_loss = -254.1 (0.78)\n",
            "epoch: 17539, dloss = 1.1, gloss = -2595.7, total_loss = -259.0 (0.70)\n",
            "epoch: 17540, dloss = 1.7, gloss = -2513.6, total_loss = -250.5 (0.68)\n",
            "epoch: 17541, dloss = 2.3, gloss = -2504.5, total_loss = -249.3 (0.69)\n",
            "epoch: 17542, dloss = 2.7, gloss = -2544.4, total_loss = -253.1 (0.72)\n",
            "epoch: 17543, dloss = -0.7, gloss = -2525.6, total_loss = -252.9 (0.70)\n",
            "epoch: 17544, dloss = 0.3, gloss = -2575.8, total_loss = -257.4 (0.68)\n",
            "epoch: 17545, dloss = 3.2, gloss = -2529.1, total_loss = -251.3 (0.69)\n",
            "epoch: 17546, dloss = 1.6, gloss = -2569.0, total_loss = -256.1 (0.70)\n",
            "epoch: 17547, dloss = -0.2, gloss = -2511.6, total_loss = -251.3 (0.67)\n",
            "epoch: 17548, dloss = 3.9, gloss = -2569.0, total_loss = -254.9 (0.69)\n",
            "epoch: 17549, dloss = 1.2, gloss = -2540.7, total_loss = -253.5 (0.68)\n",
            "epoch: 17550, dloss = 5.9, gloss = -2577.6, total_loss = -254.8 (0.72)\n",
            "epoch: 17551, dloss = 4.1, gloss = -2525.3, total_loss = -250.5 (1.00)\n",
            "epoch: 17552, dloss = 0.9, gloss = -2489.3, total_loss = -248.5 (1.13)\n",
            "epoch: 17553, dloss = 2.9, gloss = -2517.0, total_loss = -250.3 (0.78)\n",
            "epoch: 17554, dloss = 3.3, gloss = -2488.7, total_loss = -247.2 (0.69)\n",
            "epoch: 17555, dloss = 2.9, gloss = -2531.2, total_loss = -251.7 (0.67)\n",
            "epoch: 17556, dloss = -0.9, gloss = -2510.9, total_loss = -251.5 (0.67)\n",
            "epoch: 17557, dloss = -1.3, gloss = -2480.2, total_loss = -248.7 (0.67)\n",
            "epoch: 17558, dloss = 2.8, gloss = -2547.8, total_loss = -253.4 (0.67)\n",
            "epoch: 17559, dloss = -0.2, gloss = -2534.2, total_loss = -253.5 (0.68)\n",
            "epoch: 17560, dloss = -0.5, gloss = -2492.2, total_loss = -249.5 (0.67)\n",
            "epoch: 17561, dloss = 0.8, gloss = -2462.5, total_loss = -245.9 (0.67)\n",
            "epoch: 17562, dloss = 2.2, gloss = -2500.4, total_loss = -248.9 (0.69)\n",
            "epoch: 17563, dloss = -5.5, gloss = -2509.5, total_loss = -253.7 (0.71)\n",
            "epoch: 17564, dloss = 2.7, gloss = -2497.6, total_loss = -248.4 (0.69)\n",
            "epoch: 17565, dloss = 4.4, gloss = -2524.2, total_loss = -250.2 (0.68)\n",
            "epoch: 17566, dloss = -1.9, gloss = -2559.6, total_loss = -256.9 (0.71)\n",
            "epoch: 17567, dloss = -1.3, gloss = -2539.0, total_loss = -254.6 (1.26)\n",
            "epoch: 17568, dloss = 2.3, gloss = -2550.2, total_loss = -253.8 (0.70)\n",
            "epoch: 17569, dloss = 4.9, gloss = -2509.9, total_loss = -248.5 (0.71)\n",
            "epoch: 17570, dloss = 0.6, gloss = -2595.9, total_loss = -259.3 (0.68)\n",
            "epoch: 17571, dloss = 2.8, gloss = -2563.9, total_loss = -255.0 (0.70)\n",
            "epoch: 17572, dloss = 3.1, gloss = -2603.5, total_loss = -258.8 (0.77)\n",
            "epoch: 17573, dloss = 4.8, gloss = -2581.3, total_loss = -255.7 (1.01)\n",
            "epoch: 17574, dloss = 3.9, gloss = -2547.3, total_loss = -252.8 (0.71)\n",
            "epoch: 17575, dloss = 2.9, gloss = -2605.9, total_loss = -259.1 (0.67)\n",
            "epoch: 17576, dloss = 3.1, gloss = -2512.6, total_loss = -249.7 (0.79)\n",
            "epoch: 17577, dloss = 3.2, gloss = -2516.8, total_loss = -250.1 (0.96)\n",
            "epoch: 17578, dloss = 5.7, gloss = -2526.0, total_loss = -249.7 (0.68)\n",
            "epoch: 17579, dloss = 1.5, gloss = -2643.3, total_loss = -263.6 (0.69)\n",
            "epoch: 17580, dloss = -0.8, gloss = -2600.7, total_loss = -260.5 (0.72)\n",
            "epoch: 17581, dloss = 1.1, gloss = -2616.3, total_loss = -261.1 (0.92)\n",
            "epoch: 17582, dloss = 3.1, gloss = -2537.0, total_loss = -252.2 (0.81)\n",
            "epoch: 17583, dloss = 0.4, gloss = -2500.9, total_loss = -249.9 (0.69)\n",
            "epoch: 17584, dloss = 2.8, gloss = -2527.1, total_loss = -251.3 (0.69)\n",
            "epoch: 17585, dloss = 4.6, gloss = -2550.8, total_loss = -252.8 (0.70)\n",
            "epoch: 17586, dloss = 0.5, gloss = -2533.5, total_loss = -253.1 (0.71)\n",
            "epoch: 17587, dloss = 0.6, gloss = -2580.4, total_loss = -257.7 (0.72)\n",
            "epoch: 17588, dloss = 4.0, gloss = -2576.4, total_loss = -255.7 (0.70)\n",
            "epoch: 17589, dloss = 3.6, gloss = -2543.2, total_loss = -252.5 (0.69)\n",
            "epoch: 17590, dloss = 4.6, gloss = -2520.1, total_loss = -249.7 (0.69)\n",
            "epoch: 17591, dloss = 5.7, gloss = -2537.4, total_loss = -250.9 (0.70)\n",
            "epoch: 17592, dloss = 5.3, gloss = -2540.0, total_loss = -251.3 (0.69)\n",
            "epoch: 17593, dloss = 4.9, gloss = -2547.0, total_loss = -252.2 (0.69)\n",
            "epoch: 17594, dloss = 2.3, gloss = -2545.6, total_loss = -253.4 (0.70)\n",
            "epoch: 17595, dloss = 5.1, gloss = -2602.8, total_loss = -257.7 (0.67)\n",
            "epoch: 17596, dloss = 4.4, gloss = -2550.7, total_loss = -252.9 (0.82)\n",
            "epoch: 17597, dloss = 4.6, gloss = -2523.4, total_loss = -250.0 (0.89)\n",
            "epoch: 17598, dloss = 3.5, gloss = -2540.7, total_loss = -252.3 (0.69)\n",
            "epoch: 17599, dloss = 1.3, gloss = -2534.2, total_loss = -252.8 (0.69)\n",
            "epoch: 17600, dloss = 1.7, gloss = -2487.1, total_loss = -247.9 (0.69)\n",
            "epoch: 17601, dloss = 3.6, gloss = -2503.0, total_loss = -248.5 (0.70)\n",
            "epoch: 17602, dloss = 2.3, gloss = -2491.9, total_loss = -248.0 (0.68)\n",
            "epoch: 17603, dloss = 1.8, gloss = -2486.6, total_loss = -247.8 (0.74)\n",
            "epoch: 17604, dloss = 3.3, gloss = -2539.1, total_loss = -252.2 (0.68)\n",
            "epoch: 17605, dloss = 1.1, gloss = -2564.0, total_loss = -255.9 (0.70)\n",
            "epoch: 17606, dloss = 3.5, gloss = -2560.6, total_loss = -254.3 (0.70)\n",
            "epoch: 17607, dloss = 2.1, gloss = -2539.8, total_loss = -252.9 (0.69)\n",
            "epoch: 17608, dloss = 3.8, gloss = -2521.2, total_loss = -250.2 (0.69)\n",
            "epoch: 17609, dloss = 0.4, gloss = -2459.1, total_loss = -245.7 (0.70)\n",
            "epoch: 17610, dloss = 3.8, gloss = -2469.5, total_loss = -245.0 (0.68)\n",
            "epoch: 17611, dloss = 1.3, gloss = -2512.8, total_loss = -250.6 (0.66)\n",
            "epoch: 17612, dloss = 2.3, gloss = -2558.0, total_loss = -254.6 (1.01)\n",
            "epoch: 17613, dloss = 0.4, gloss = -2568.3, total_loss = -256.6 (0.71)\n",
            "epoch: 17614, dloss = 6.0, gloss = -2575.5, total_loss = -254.5 (0.68)\n",
            "epoch: 17615, dloss = 2.5, gloss = -2556.1, total_loss = -254.3 (0.69)\n",
            "epoch: 17616, dloss = 4.3, gloss = -2544.5, total_loss = -252.3 (0.70)\n",
            "epoch: 17617, dloss = 2.6, gloss = -2529.6, total_loss = -251.7 (0.69)\n",
            "epoch: 17618, dloss = 4.3, gloss = -2568.1, total_loss = -254.7 (0.67)\n",
            "epoch: 17619, dloss = 4.9, gloss = -2535.1, total_loss = -251.1 (0.71)\n",
            "epoch: 17620, dloss = 0.6, gloss = -2489.7, total_loss = -248.7 (0.69)\n",
            "epoch: 17621, dloss = 3.0, gloss = -2518.3, total_loss = -250.3 (0.69)\n",
            "epoch: 17622, dloss = 2.4, gloss = -2484.5, total_loss = -247.3 (0.68)\n",
            "epoch: 17623, dloss = 2.7, gloss = -2484.2, total_loss = -247.1 (0.69)\n",
            "epoch: 17624, dloss = 2.9, gloss = -2500.1, total_loss = -248.5 (0.70)\n",
            "epoch: 17625, dloss = 1.1, gloss = -2514.1, total_loss = -250.9 (0.69)\n",
            "epoch: 17626, dloss = 0.8, gloss = -2568.6, total_loss = -256.4 (0.71)\n",
            "epoch: 17627, dloss = 3.3, gloss = -2559.9, total_loss = -254.3 (0.87)\n",
            "epoch: 17628, dloss = 1.2, gloss = -2595.9, total_loss = -259.0 (0.89)\n",
            "epoch: 17629, dloss = 3.3, gloss = -2578.0, total_loss = -256.1 (0.69)\n",
            "epoch: 17630, dloss = 1.4, gloss = -2485.3, total_loss = -247.8 (0.69)\n",
            "epoch: 17631, dloss = 3.1, gloss = -2532.3, total_loss = -251.7 (0.69)\n",
            "epoch: 17632, dloss = 0.9, gloss = -2494.5, total_loss = -249.0 (0.70)\n",
            "epoch: 17633, dloss = 2.5, gloss = -2498.8, total_loss = -248.7 (0.70)\n",
            "epoch: 17634, dloss = 4.5, gloss = -2561.0, total_loss = -253.9 (0.70)\n",
            "epoch: 17635, dloss = 1.0, gloss = -2527.7, total_loss = -252.3 (0.69)\n",
            "epoch: 17636, dloss = 1.7, gloss = -2534.2, total_loss = -252.6 (0.69)\n",
            "epoch: 17637, dloss = -5.1, gloss = -2537.9, total_loss = -256.3 (0.71)\n",
            "epoch: 17638, dloss = 2.2, gloss = -2631.1, total_loss = -262.0 (0.69)\n",
            "epoch: 17639, dloss = 1.1, gloss = -2553.7, total_loss = -254.8 (0.69)\n",
            "epoch: 17640, dloss = -1.6, gloss = -2624.7, total_loss = -263.2 (0.69)\n",
            "epoch: 17641, dloss = 3.2, gloss = -2594.7, total_loss = -257.9 (0.69)\n",
            "epoch: 17642, dloss = -0.3, gloss = -2556.4, total_loss = -255.8 (0.71)\n",
            "epoch: 17643, dloss = -3.9, gloss = -2617.3, total_loss = -263.7 (1.02)\n",
            "epoch: 17644, dloss = 2.4, gloss = -2650.0, total_loss = -263.8 (0.67)\n",
            "epoch: 17645, dloss = 4.2, gloss = -2536.7, total_loss = -251.6 (0.66)\n",
            "epoch: 17646, dloss = 4.1, gloss = -2516.2, total_loss = -249.6 (0.72)\n",
            "epoch: 17647, dloss = 2.1, gloss = -2534.3, total_loss = -252.4 (0.69)\n",
            "epoch: 17648, dloss = 4.8, gloss = -2535.5, total_loss = -251.1 (0.69)\n",
            "epoch: 17649, dloss = 1.9, gloss = -2573.2, total_loss = -256.4 (0.69)\n",
            "epoch: 17650, dloss = 2.3, gloss = -2560.1, total_loss = -254.9 (0.68)\n",
            "epoch: 17651, dloss = 3.0, gloss = -2536.4, total_loss = -252.1 (0.71)\n",
            "epoch: 17652, dloss = 3.3, gloss = -2566.9, total_loss = -255.0 (0.68)\n",
            "epoch: 17653, dloss = -0.7, gloss = -2530.8, total_loss = -253.4 (0.69)\n",
            "epoch: 17654, dloss = 3.8, gloss = -2644.3, total_loss = -262.5 (0.70)\n",
            "epoch: 17655, dloss = 2.4, gloss = -2584.4, total_loss = -257.2 (0.66)\n",
            "epoch: 17656, dloss = 1.2, gloss = -2511.4, total_loss = -250.6 (0.69)\n",
            "epoch: 17657, dloss = 2.6, gloss = -2501.4, total_loss = -248.8 (0.70)\n",
            "epoch: 17658, dloss = 1.4, gloss = -2474.0, total_loss = -246.7 (0.87)\n",
            "epoch: 17659, dloss = 1.0, gloss = -2486.8, total_loss = -248.2 (0.87)\n",
            "epoch: 17660, dloss = 6.2, gloss = -2497.9, total_loss = -246.7 (0.70)\n",
            "epoch: 17661, dloss = 3.5, gloss = -2519.3, total_loss = -250.2 (0.69)\n",
            "epoch: 17662, dloss = 0.9, gloss = -2560.6, total_loss = -255.6 (0.67)\n",
            "epoch: 17663, dloss = 3.9, gloss = -2584.9, total_loss = -256.5 (0.70)\n",
            "epoch: 17664, dloss = 3.2, gloss = -2601.1, total_loss = -258.5 (0.69)\n",
            "epoch: 17665, dloss = -0.2, gloss = -2526.6, total_loss = -252.8 (0.68)\n",
            "epoch: 17666, dloss = 2.7, gloss = -2554.5, total_loss = -254.1 (0.69)\n",
            "epoch: 17667, dloss = 3.7, gloss = -2566.5, total_loss = -254.8 (0.69)\n",
            "epoch: 17668, dloss = 2.7, gloss = -2587.6, total_loss = -257.4 (0.67)\n",
            "epoch: 17669, dloss = 3.0, gloss = -2538.8, total_loss = -252.4 (0.70)\n",
            "epoch: 17670, dloss = 4.1, gloss = -2554.8, total_loss = -253.4 (0.71)\n",
            "epoch: 17671, dloss = 3.8, gloss = -2537.3, total_loss = -251.8 (0.68)\n",
            "epoch: 17672, dloss = 1.7, gloss = -2585.7, total_loss = -257.7 (0.68)\n",
            "epoch: 17673, dloss = 1.3, gloss = -2538.8, total_loss = -253.2 (1.26)\n",
            "epoch: 17674, dloss = 3.3, gloss = -2580.4, total_loss = -256.4 (0.70)\n",
            "epoch: 17675, dloss = 5.2, gloss = -2657.3, total_loss = -263.1 (0.70)\n",
            "epoch: 17676, dloss = 0.9, gloss = -2591.4, total_loss = -258.7 (0.68)\n",
            "epoch: 17677, dloss = 3.5, gloss = -2557.8, total_loss = -254.1 (0.70)\n",
            "epoch: 17678, dloss = 4.2, gloss = -2558.4, total_loss = -253.8 (0.69)\n",
            "epoch: 17679, dloss = 2.1, gloss = -2511.6, total_loss = -250.1 (0.69)\n",
            "epoch: 17680, dloss = 1.9, gloss = -2600.8, total_loss = -259.1 (0.68)\n",
            "epoch: 17681, dloss = 2.2, gloss = -2586.2, total_loss = -257.5 (0.68)\n",
            "epoch: 17682, dloss = 6.3, gloss = -2563.8, total_loss = -253.2 (0.67)\n",
            "epoch: 17683, dloss = 4.2, gloss = -2603.8, total_loss = -258.3 (0.68)\n",
            "epoch: 17684, dloss = 3.0, gloss = -2558.7, total_loss = -254.4 (0.66)\n",
            "epoch: 17685, dloss = 2.9, gloss = -2531.6, total_loss = -251.7 (0.68)\n",
            "epoch: 17686, dloss = 1.2, gloss = -2570.1, total_loss = -256.4 (0.72)\n",
            "epoch: 17687, dloss = 2.2, gloss = -2523.9, total_loss = -251.3 (0.71)\n",
            "epoch: 17688, dloss = 5.0, gloss = -2503.3, total_loss = -247.8 (0.87)\n",
            "epoch: 17689, dloss = 2.0, gloss = -2513.6, total_loss = -250.4 (0.89)\n",
            "epoch: 17690, dloss = 2.8, gloss = -2518.2, total_loss = -250.4 (0.67)\n",
            "epoch: 17691, dloss = 2.2, gloss = -2530.4, total_loss = -252.0 (0.67)\n",
            "epoch: 17692, dloss = -1.0, gloss = -2545.0, total_loss = -255.0 (0.71)\n",
            "epoch: 17693, dloss = 4.3, gloss = -2521.1, total_loss = -250.0 (0.69)\n",
            "epoch: 17694, dloss = 1.0, gloss = -2519.8, total_loss = -251.5 (0.68)\n",
            "epoch: 17695, dloss = 5.4, gloss = -2539.2, total_loss = -251.2 (0.69)\n",
            "epoch: 17696, dloss = 11.8, gloss = -2546.4, total_loss = -248.8 (0.69)\n",
            "epoch: 17697, dloss = 3.9, gloss = -2515.2, total_loss = -249.6 (0.69)\n",
            "epoch: 17698, dloss = 3.0, gloss = -2521.3, total_loss = -250.6 (0.69)\n",
            "epoch: 17699, dloss = 0.7, gloss = -2490.4, total_loss = -248.7 (0.68)\n",
            "epoch: 17700, dloss = 1.8, gloss = -2491.1, total_loss = -248.2 (0.68)\n",
            "epoch: 17701, dloss = 2.3, gloss = -2511.4, total_loss = -250.0 (0.69)\n",
            "epoch: 17702, dloss = 5.7, gloss = -2479.1, total_loss = -245.1 (0.67)\n",
            "epoch: 17703, dloss = 3.9, gloss = -2584.4, total_loss = -256.5 (0.69)\n",
            "epoch: 17704, dloss = 2.9, gloss = -2499.4, total_loss = -248.5 (1.01)\n",
            "epoch: 17705, dloss = 1.6, gloss = -2513.0, total_loss = -250.5 (0.76)\n",
            "epoch: 17706, dloss = 2.7, gloss = -2523.3, total_loss = -251.0 (0.70)\n",
            "epoch: 17707, dloss = 2.5, gloss = -2581.7, total_loss = -256.9 (0.69)\n",
            "epoch: 17708, dloss = 1.1, gloss = -2622.0, total_loss = -261.6 (0.68)\n",
            "epoch: 17709, dloss = 6.2, gloss = -2570.8, total_loss = -254.0 (0.68)\n",
            "epoch: 17710, dloss = 3.5, gloss = -2572.0, total_loss = -255.4 (1.26)\n",
            "epoch: 17711, dloss = 3.6, gloss = -2571.4, total_loss = -255.4 (0.67)\n",
            "epoch: 17712, dloss = 3.3, gloss = -2547.1, total_loss = -253.1 (0.71)\n",
            "epoch: 17713, dloss = 2.1, gloss = -2556.6, total_loss = -254.6 (0.70)\n",
            "epoch: 17714, dloss = 1.3, gloss = -2588.1, total_loss = -258.1 (0.67)\n",
            "epoch: 17715, dloss = 3.2, gloss = -2560.4, total_loss = -254.4 (0.66)\n",
            "epoch: 17716, dloss = 3.1, gloss = -2620.7, total_loss = -260.5 (0.70)\n",
            "epoch: 17717, dloss = 4.2, gloss = -2628.7, total_loss = -260.8 (0.68)\n",
            "epoch: 17718, dloss = 5.7, gloss = -2623.0, total_loss = -259.5 (0.80)\n",
            "epoch: 17719, dloss = 5.0, gloss = -2621.5, total_loss = -259.7 (0.99)\n",
            "epoch: 17720, dloss = 3.6, gloss = -2645.9, total_loss = -262.8 (0.71)\n",
            "epoch: 17721, dloss = 1.4, gloss = -2618.2, total_loss = -261.1 (0.69)\n",
            "epoch: 17722, dloss = 2.3, gloss = -2579.9, total_loss = -256.8 (0.71)\n",
            "epoch: 17723, dloss = 1.4, gloss = -2598.0, total_loss = -259.1 (0.71)\n",
            "epoch: 17724, dloss = 1.2, gloss = -2541.7, total_loss = -253.6 (0.69)\n",
            "epoch: 17725, dloss = 2.6, gloss = -2568.7, total_loss = -255.6 (0.71)\n",
            "epoch: 17726, dloss = 3.5, gloss = -2569.0, total_loss = -255.1 (0.69)\n",
            "epoch: 17727, dloss = 2.7, gloss = -2554.6, total_loss = -254.1 (0.68)\n",
            "epoch: 17728, dloss = 0.6, gloss = -2520.5, total_loss = -251.7 (0.70)\n",
            "epoch: 17729, dloss = 0.1, gloss = -2476.3, total_loss = -247.6 (0.70)\n",
            "epoch: 17730, dloss = -2.1, gloss = -2507.9, total_loss = -251.8 (0.70)\n",
            "epoch: 17731, dloss = 1.5, gloss = -2595.3, total_loss = -258.8 (0.68)\n",
            "epoch: 17732, dloss = 3.2, gloss = -2524.2, total_loss = -250.8 (0.69)\n",
            "epoch: 17733, dloss = 4.7, gloss = -2578.6, total_loss = -255.5 (0.68)\n",
            "epoch: 17734, dloss = 4.9, gloss = -2518.7, total_loss = -249.4 (1.26)\n",
            "epoch: 17735, dloss = 3.1, gloss = -2519.7, total_loss = -250.4 (0.68)\n",
            "epoch: 17736, dloss = -0.2, gloss = -2460.9, total_loss = -246.2 (0.67)\n",
            "epoch: 17737, dloss = 1.8, gloss = -2460.8, total_loss = -245.2 (0.65)\n",
            "epoch: 17738, dloss = -0.5, gloss = -2463.6, total_loss = -246.6 (0.67)\n",
            "epoch: 17739, dloss = 0.5, gloss = -2489.7, total_loss = -248.7 (0.70)\n",
            "epoch: 17740, dloss = 2.2, gloss = -2484.7, total_loss = -247.4 (0.68)\n",
            "epoch: 17741, dloss = 3.3, gloss = -2484.3, total_loss = -246.8 (0.68)\n",
            "epoch: 17742, dloss = 1.7, gloss = -2416.7, total_loss = -240.8 (0.68)\n",
            "epoch: 17743, dloss = 2.0, gloss = -2389.5, total_loss = -238.0 (0.68)\n",
            "epoch: 17744, dloss = 4.4, gloss = -2419.1, total_loss = -239.7 (0.65)\n",
            "epoch: 17745, dloss = 21.7, gloss = -2483.8, total_loss = -237.5 (0.67)\n",
            "epoch: 17746, dloss = 4.2, gloss = -2514.4, total_loss = -249.4 (0.69)\n",
            "epoch: 17747, dloss = 4.1, gloss = -2505.6, total_loss = -248.5 (0.66)\n",
            "epoch: 17748, dloss = 3.5, gloss = -2531.5, total_loss = -251.4 (0.69)\n",
            "epoch: 17749, dloss = 2.2, gloss = -2518.2, total_loss = -250.7 (0.87)\n",
            "epoch: 17750, dloss = 4.2, gloss = -2516.1, total_loss = -249.5 (0.87)\n",
            "epoch: 17751, dloss = 0.5, gloss = -2458.8, total_loss = -245.6 (0.70)\n",
            "epoch: 17752, dloss = 1.9, gloss = -2547.6, total_loss = -253.8 (0.68)\n",
            "epoch: 17753, dloss = 3.8, gloss = -2458.4, total_loss = -244.0 (0.70)\n",
            "epoch: 17754, dloss = 3.0, gloss = -2464.3, total_loss = -244.9 (0.69)\n",
            "epoch: 17755, dloss = 4.8, gloss = -2465.1, total_loss = -244.1 (0.67)\n",
            "epoch: 17756, dloss = 1.7, gloss = -2521.2, total_loss = -251.3 (0.67)\n",
            "epoch: 17757, dloss = 3.7, gloss = -2465.1, total_loss = -244.7 (0.64)\n",
            "epoch: 17758, dloss = 3.5, gloss = -2548.3, total_loss = -253.1 (0.66)\n",
            "epoch: 17759, dloss = 4.3, gloss = -2541.6, total_loss = -252.0 (0.68)\n",
            "epoch: 17760, dloss = 4.5, gloss = -2570.7, total_loss = -254.8 (0.67)\n",
            "epoch: 17761, dloss = -0.0, gloss = -2580.1, total_loss = -258.0 (0.70)\n",
            "epoch: 17762, dloss = 2.2, gloss = -2593.2, total_loss = -258.2 (0.68)\n",
            "epoch: 17763, dloss = 7.7, gloss = -2569.2, total_loss = -253.1 (0.68)\n",
            "epoch: 17764, dloss = 0.8, gloss = -2580.3, total_loss = -257.6 (0.68)\n",
            "epoch: 17765, dloss = 2.3, gloss = -2607.5, total_loss = -259.6 (0.93)\n",
            "epoch: 17766, dloss = 2.4, gloss = -2581.6, total_loss = -257.0 (0.79)\n",
            "epoch: 17767, dloss = 4.1, gloss = -2619.0, total_loss = -259.8 (0.67)\n",
            "epoch: 17768, dloss = 4.3, gloss = -2571.1, total_loss = -254.9 (0.74)\n",
            "epoch: 17769, dloss = 5.2, gloss = -2516.4, total_loss = -249.1 (0.69)\n",
            "epoch: 17770, dloss = 0.3, gloss = -2548.4, total_loss = -254.7 (0.68)\n",
            "epoch: 17771, dloss = 2.8, gloss = -2505.0, total_loss = -249.1 (0.67)\n",
            "epoch: 17772, dloss = 1.1, gloss = -2533.8, total_loss = -252.8 (0.69)\n",
            "epoch: 17773, dloss = -0.7, gloss = -2485.5, total_loss = -248.9 (0.68)\n",
            "epoch: 17774, dloss = 3.2, gloss = -2566.1, total_loss = -255.0 (0.68)\n",
            "epoch: 17775, dloss = 2.5, gloss = -2473.8, total_loss = -246.1 (0.68)\n",
            "epoch: 17776, dloss = 3.6, gloss = -2467.2, total_loss = -244.9 (0.68)\n",
            "epoch: 17777, dloss = 1.5, gloss = -2496.5, total_loss = -248.9 (0.68)\n",
            "epoch: 17778, dloss = 20.7, gloss = -2506.0, total_loss = -240.3 (0.68)\n",
            "epoch: 17779, dloss = 2.7, gloss = -2511.7, total_loss = -249.8 (0.68)\n",
            "epoch: 17780, dloss = 1.9, gloss = -2508.3, total_loss = -249.9 (1.26)\n",
            "epoch: 17781, dloss = 3.5, gloss = -2531.8, total_loss = -251.4 (0.69)\n",
            "epoch: 17782, dloss = 4.2, gloss = -2505.5, total_loss = -248.5 (0.69)\n",
            "epoch: 17783, dloss = 5.2, gloss = -2548.7, total_loss = -252.3 (0.69)\n",
            "epoch: 17784, dloss = 3.0, gloss = -2609.2, total_loss = -259.4 (0.70)\n",
            "epoch: 17785, dloss = 2.1, gloss = -2603.5, total_loss = -259.3 (0.67)\n",
            "epoch: 17786, dloss = 2.2, gloss = -2563.9, total_loss = -255.3 (0.67)\n",
            "epoch: 17787, dloss = 3.2, gloss = -2583.7, total_loss = -256.8 (0.68)\n",
            "epoch: 17788, dloss = 2.9, gloss = -2589.9, total_loss = -257.6 (0.67)\n",
            "epoch: 17789, dloss = 2.2, gloss = -2611.2, total_loss = -260.0 (0.67)\n",
            "epoch: 17790, dloss = -0.1, gloss = -2565.5, total_loss = -256.6 (0.69)\n",
            "epoch: 17791, dloss = 2.9, gloss = -2618.0, total_loss = -260.3 (0.67)\n",
            "epoch: 17792, dloss = 2.3, gloss = -2633.7, total_loss = -262.2 (0.68)\n",
            "epoch: 17793, dloss = 3.8, gloss = -2592.9, total_loss = -257.4 (0.69)\n",
            "epoch: 17794, dloss = 35.5, gloss = -2645.6, total_loss = -246.8 (0.68)\n",
            "epoch: 17795, dloss = 1.6, gloss = -2630.5, total_loss = -262.2 (0.81)\n",
            "epoch: 17796, dloss = 2.6, gloss = -2677.1, total_loss = -266.4 (0.92)\n",
            "epoch: 17797, dloss = 2.7, gloss = -2607.7, total_loss = -259.4 (0.68)\n",
            "epoch: 17798, dloss = 5.2, gloss = -2541.0, total_loss = -251.5 (0.69)\n",
            "epoch: 17799, dloss = 4.9, gloss = -2544.4, total_loss = -252.0 (0.68)\n",
            "epoch: 17800, dloss = 1.9, gloss = -2541.9, total_loss = -253.2 (0.67)\n",
            "epoch: 17801, dloss = -1.7, gloss = -2513.7, total_loss = -252.2 (0.69)\n",
            "epoch: 17802, dloss = 6.1, gloss = -2526.7, total_loss = -249.6 (0.69)\n",
            "epoch: 17803, dloss = 3.2, gloss = -2537.3, total_loss = -252.1 (0.70)\n",
            "epoch: 17804, dloss = 3.9, gloss = -2545.1, total_loss = -252.6 (0.67)\n",
            "epoch: 17805, dloss = 3.4, gloss = -2547.3, total_loss = -253.0 (0.68)\n",
            "epoch: 17806, dloss = 4.1, gloss = -2623.9, total_loss = -260.3 (0.67)\n",
            "epoch: 17807, dloss = 3.1, gloss = -2633.3, total_loss = -261.8 (0.70)\n",
            "epoch: 17808, dloss = 3.9, gloss = -2639.8, total_loss = -262.0 (0.69)\n",
            "epoch: 17809, dloss = 5.6, gloss = -2636.7, total_loss = -260.8 (0.70)\n",
            "epoch: 17810, dloss = 4.4, gloss = -2584.7, total_loss = -256.3 (0.66)\n",
            "epoch: 17811, dloss = 5.1, gloss = -2565.9, total_loss = -254.0 (0.94)\n",
            "epoch: 17812, dloss = 2.3, gloss = -2547.3, total_loss = -253.6 (0.81)\n",
            "epoch: 17813, dloss = 3.1, gloss = -2552.6, total_loss = -253.7 (0.67)\n",
            "epoch: 17814, dloss = 4.0, gloss = -2565.3, total_loss = -254.5 (0.70)\n",
            "epoch: 17815, dloss = 2.3, gloss = -2584.6, total_loss = -257.3 (0.67)\n",
            "epoch: 17816, dloss = 3.5, gloss = -2582.8, total_loss = -256.5 (0.71)\n",
            "epoch: 17817, dloss = 5.2, gloss = -2631.8, total_loss = -260.6 (0.69)\n",
            "epoch: 17818, dloss = 4.9, gloss = -2631.8, total_loss = -260.7 (0.71)\n",
            "epoch: 17819, dloss = 2.4, gloss = -2618.5, total_loss = -260.6 (0.70)\n",
            "epoch: 17820, dloss = 2.0, gloss = -2635.1, total_loss = -262.5 (0.69)\n",
            "epoch: 17821, dloss = 0.3, gloss = -2624.8, total_loss = -262.3 (0.71)\n",
            "epoch: 17822, dloss = 0.8, gloss = -2625.5, total_loss = -262.2 (0.68)\n",
            "epoch: 17823, dloss = 21.9, gloss = -2620.9, total_loss = -251.2 (0.69)\n",
            "epoch: 17824, dloss = 2.3, gloss = -2607.8, total_loss = -259.6 (0.68)\n",
            "epoch: 17825, dloss = 1.2, gloss = -2612.8, total_loss = -260.7 (0.69)\n",
            "epoch: 17826, dloss = 4.7, gloss = -2624.9, total_loss = -260.1 (0.74)\n",
            "epoch: 17827, dloss = 3.7, gloss = -2598.7, total_loss = -258.0 (0.98)\n",
            "epoch: 17828, dloss = 2.8, gloss = -2584.7, total_loss = -257.0 (0.71)\n",
            "epoch: 17829, dloss = 4.6, gloss = -2609.2, total_loss = -258.6 (0.68)\n",
            "epoch: 17830, dloss = 2.7, gloss = -2573.6, total_loss = -256.0 (0.67)\n",
            "epoch: 17831, dloss = 4.9, gloss = -2607.9, total_loss = -258.3 (0.68)\n",
            "epoch: 17832, dloss = 2.8, gloss = -2586.4, total_loss = -257.3 (0.70)\n",
            "epoch: 17833, dloss = 3.7, gloss = -2602.2, total_loss = -258.4 (0.69)\n",
            "epoch: 17834, dloss = 4.6, gloss = -2550.9, total_loss = -252.8 (0.68)\n",
            "epoch: 17835, dloss = 4.2, gloss = -2533.8, total_loss = -251.3 (0.69)\n",
            "epoch: 17836, dloss = 2.0, gloss = -2541.9, total_loss = -253.2 (0.69)\n",
            "epoch: 17837, dloss = 2.8, gloss = -2540.2, total_loss = -252.6 (0.68)\n",
            "epoch: 17838, dloss = 4.0, gloss = -2522.1, total_loss = -250.2 (0.89)\n",
            "epoch: 17839, dloss = 1.8, gloss = -2526.8, total_loss = -251.8 (0.87)\n",
            "epoch: 17840, dloss = 4.3, gloss = -2570.1, total_loss = -254.9 (0.69)\n",
            "epoch: 17841, dloss = 5.1, gloss = -2572.7, total_loss = -254.7 (0.77)\n",
            "epoch: 17842, dloss = 3.3, gloss = -2598.7, total_loss = -258.2 (0.95)\n",
            "epoch: 17843, dloss = 2.0, gloss = -2614.4, total_loss = -260.4 (0.68)\n",
            "epoch: 17844, dloss = 2.1, gloss = -2567.7, total_loss = -255.7 (0.68)\n",
            "epoch: 17845, dloss = 2.3, gloss = -2547.2, total_loss = -253.6 (0.85)\n",
            "epoch: 17846, dloss = 1.8, gloss = -2550.6, total_loss = -254.2 (0.88)\n",
            "epoch: 17847, dloss = -2.5, gloss = -2536.3, total_loss = -254.9 (0.67)\n",
            "epoch: 17848, dloss = 3.1, gloss = -2573.0, total_loss = -255.7 (0.68)\n",
            "epoch: 17849, dloss = 0.2, gloss = -2593.6, total_loss = -259.3 (0.67)\n",
            "epoch: 17850, dloss = 2.6, gloss = -2596.9, total_loss = -258.4 (0.69)\n",
            "epoch: 17851, dloss = 0.4, gloss = -2588.7, total_loss = -258.7 (0.67)\n",
            "epoch: 17852, dloss = 5.8, gloss = -2618.5, total_loss = -259.0 (0.70)\n",
            "epoch: 17853, dloss = 3.3, gloss = -2635.6, total_loss = -261.9 (0.69)\n",
            "epoch: 17854, dloss = -0.3, gloss = -2613.0, total_loss = -261.5 (0.69)\n",
            "epoch: 17855, dloss = 1.3, gloss = -2614.3, total_loss = -260.8 (0.70)\n",
            "epoch: 17856, dloss = 2.3, gloss = -2575.9, total_loss = -256.4 (0.78)\n",
            "epoch: 17857, dloss = 2.4, gloss = -2578.9, total_loss = -256.7 (0.96)\n",
            "epoch: 17858, dloss = 0.6, gloss = -2570.6, total_loss = -256.7 (0.70)\n",
            "epoch: 17859, dloss = 1.8, gloss = -2558.0, total_loss = -254.9 (0.68)\n",
            "epoch: 17860, dloss = 1.8, gloss = -2571.4, total_loss = -256.2 (0.79)\n",
            "epoch: 17861, dloss = -0.1, gloss = -2592.2, total_loss = -259.3 (1.04)\n",
            "epoch: 17862, dloss = -0.0, gloss = -2608.8, total_loss = -260.9 (0.68)\n",
            "epoch: 17863, dloss = 2.3, gloss = -2588.3, total_loss = -257.7 (0.67)\n",
            "epoch: 17864, dloss = 1.5, gloss = -2579.4, total_loss = -257.2 (0.68)\n",
            "epoch: 17865, dloss = 6.1, gloss = -2619.1, total_loss = -258.9 (0.68)\n",
            "epoch: 17866, dloss = 2.5, gloss = -2597.7, total_loss = -258.5 (0.67)\n",
            "epoch: 17867, dloss = 5.3, gloss = -2625.6, total_loss = -259.9 (0.68)\n",
            "epoch: 17868, dloss = 3.9, gloss = -2622.6, total_loss = -260.3 (0.68)\n",
            "epoch: 17869, dloss = 5.6, gloss = -2647.1, total_loss = -261.9 (0.67)\n",
            "epoch: 17870, dloss = 42.5, gloss = -2616.3, total_loss = -240.4 (0.69)\n",
            "epoch: 17871, dloss = 18.8, gloss = -2616.6, total_loss = -252.2 (0.77)\n",
            "epoch: 17872, dloss = 5.6, gloss = -2665.8, total_loss = -263.8 (0.98)\n",
            "epoch: 17873, dloss = 5.1, gloss = -2679.7, total_loss = -265.4 (0.66)\n",
            "epoch: 17874, dloss = 3.4, gloss = -2652.1, total_loss = -263.5 (0.70)\n",
            "epoch: 17875, dloss = 5.2, gloss = -2648.9, total_loss = -262.3 (0.68)\n",
            "epoch: 17876, dloss = 4.4, gloss = -2644.7, total_loss = -262.2 (0.67)\n",
            "epoch: 17877, dloss = 0.8, gloss = -2650.1, total_loss = -264.6 (0.65)\n",
            "epoch: 17878, dloss = 3.9, gloss = -2638.3, total_loss = -261.9 (0.69)\n",
            "epoch: 17879, dloss = 3.1, gloss = -2668.5, total_loss = -265.3 (0.69)\n",
            "epoch: 17880, dloss = 2.1, gloss = -2632.5, total_loss = -262.2 (0.68)\n",
            "epoch: 17881, dloss = 0.8, gloss = -2620.7, total_loss = -261.7 (0.69)\n",
            "epoch: 17882, dloss = 3.5, gloss = -2573.3, total_loss = -255.6 (0.66)\n",
            "epoch: 17883, dloss = 2.2, gloss = -2562.3, total_loss = -255.1 (0.67)\n",
            "epoch: 17884, dloss = 3.1, gloss = -2608.4, total_loss = -259.3 (0.69)\n",
            "epoch: 17885, dloss = 7.5, gloss = -2620.1, total_loss = -258.2 (0.69)\n",
            "epoch: 17886, dloss = 2.6, gloss = -2599.5, total_loss = -258.7 (0.67)\n",
            "epoch: 17887, dloss = 1.2, gloss = -2545.4, total_loss = -253.9 (0.86)\n",
            "epoch: 17888, dloss = 3.3, gloss = -2522.1, total_loss = -250.6 (0.88)\n",
            "epoch: 17889, dloss = 3.6, gloss = -2546.5, total_loss = -252.8 (0.70)\n",
            "epoch: 17890, dloss = 3.2, gloss = -2549.7, total_loss = -253.4 (0.69)\n",
            "epoch: 17891, dloss = 4.2, gloss = -2575.0, total_loss = -255.4 (0.70)\n",
            "epoch: 17892, dloss = 2.4, gloss = -2591.0, total_loss = -257.9 (0.69)\n",
            "epoch: 17893, dloss = 13.5, gloss = -2564.6, total_loss = -249.7 (0.66)\n",
            "epoch: 17894, dloss = 5.1, gloss = -2590.7, total_loss = -256.5 (0.67)\n",
            "epoch: 17895, dloss = 3.6, gloss = -2639.7, total_loss = -262.1 (0.68)\n",
            "epoch: 17896, dloss = 2.9, gloss = -2570.7, total_loss = -255.6 (0.69)\n",
            "epoch: 17897, dloss = 3.1, gloss = -2565.1, total_loss = -254.9 (0.69)\n",
            "epoch: 17898, dloss = 2.8, gloss = -2572.7, total_loss = -255.9 (0.69)\n",
            "epoch: 17899, dloss = -1.7, gloss = -2566.8, total_loss = -257.5 (0.67)\n",
            "epoch: 17900, dloss = 3.7, gloss = -2587.6, total_loss = -256.9 (0.67)\n",
            "epoch: 17901, dloss = 3.5, gloss = -2546.1, total_loss = -252.9 (0.69)\n",
            "epoch: 17902, dloss = 4.8, gloss = -2559.5, total_loss = -253.6 (0.67)\n",
            "epoch: 17903, dloss = 1.6, gloss = -2540.7, total_loss = -253.3 (1.06)\n",
            "epoch: 17904, dloss = 16.9, gloss = -2620.9, total_loss = -253.6 (0.70)\n",
            "epoch: 17905, dloss = 2.9, gloss = -2599.3, total_loss = -258.5 (0.69)\n",
            "epoch: 17906, dloss = 2.2, gloss = -2526.8, total_loss = -251.6 (0.70)\n",
            "epoch: 17907, dloss = 4.8, gloss = -2550.0, total_loss = -252.6 (0.70)\n",
            "epoch: 17908, dloss = 2.8, gloss = -2569.7, total_loss = -255.6 (0.68)\n",
            "epoch: 17909, dloss = 1.0, gloss = -2613.4, total_loss = -260.8 (0.69)\n",
            "epoch: 17910, dloss = 5.0, gloss = -2578.4, total_loss = -255.4 (0.69)\n",
            "epoch: 17911, dloss = 3.1, gloss = -2594.3, total_loss = -257.9 (0.69)\n",
            "epoch: 17912, dloss = 0.8, gloss = -2566.6, total_loss = -256.3 (0.69)\n",
            "epoch: 17913, dloss = 1.0, gloss = -2571.1, total_loss = -256.6 (0.69)\n",
            "epoch: 17914, dloss = -1.8, gloss = -2527.7, total_loss = -253.7 (0.68)\n",
            "epoch: 17915, dloss = 5.3, gloss = -2573.8, total_loss = -254.7 (0.68)\n",
            "epoch: 17916, dloss = 2.6, gloss = -2501.1, total_loss = -248.8 (0.70)\n",
            "epoch: 17917, dloss = 0.9, gloss = -2481.3, total_loss = -247.7 (0.68)\n",
            "epoch: 17918, dloss = -0.4, gloss = -2549.3, total_loss = -255.1 (0.89)\n",
            "epoch: 17919, dloss = 2.2, gloss = -2608.7, total_loss = -259.8 (0.86)\n",
            "epoch: 17920, dloss = 4.3, gloss = -2562.7, total_loss = -254.1 (0.68)\n",
            "epoch: 17921, dloss = 1.1, gloss = -2577.4, total_loss = -257.2 (0.70)\n",
            "epoch: 17922, dloss = -1.8, gloss = -2558.0, total_loss = -256.7 (0.69)\n",
            "epoch: 17923, dloss = 4.1, gloss = -2535.9, total_loss = -251.5 (0.67)\n",
            "epoch: 17924, dloss = 3.5, gloss = -2575.4, total_loss = -255.8 (0.67)\n",
            "epoch: 17925, dloss = 7.9, gloss = -2545.6, total_loss = -250.6 (0.69)\n",
            "epoch: 17926, dloss = 16.0, gloss = -2553.3, total_loss = -247.3 (0.70)\n",
            "epoch: 17927, dloss = 1.8, gloss = -2588.2, total_loss = -257.9 (0.68)\n",
            "epoch: 17928, dloss = -0.3, gloss = -2547.6, total_loss = -254.9 (0.68)\n",
            "epoch: 17929, dloss = 3.7, gloss = -2568.4, total_loss = -255.0 (0.68)\n",
            "epoch: 17930, dloss = 1.2, gloss = -2490.4, total_loss = -248.4 (0.68)\n",
            "epoch: 17931, dloss = 2.1, gloss = -2508.9, total_loss = -249.8 (0.69)\n",
            "epoch: 17932, dloss = 5.7, gloss = -2513.6, total_loss = -248.5 (0.67)\n",
            "epoch: 17933, dloss = 4.2, gloss = -2505.3, total_loss = -248.4 (0.67)\n",
            "epoch: 17934, dloss = 3.6, gloss = -2542.9, total_loss = -252.5 (1.05)\n",
            "epoch: 17935, dloss = 2.1, gloss = -2517.1, total_loss = -250.7 (0.69)\n",
            "epoch: 17936, dloss = 0.8, gloss = -2570.3, total_loss = -256.6 (0.67)\n",
            "epoch: 17937, dloss = 1.5, gloss = -2607.7, total_loss = -260.0 (0.67)\n",
            "epoch: 17938, dloss = 0.8, gloss = -2624.9, total_loss = -262.1 (0.69)\n",
            "epoch: 17939, dloss = 2.9, gloss = -2589.7, total_loss = -257.5 (0.68)\n",
            "epoch: 17940, dloss = 6.5, gloss = -2676.7, total_loss = -264.4 (0.69)\n",
            "epoch: 17941, dloss = 6.5, gloss = -2776.0, total_loss = -274.4 (0.68)\n",
            "epoch: 17942, dloss = 4.3, gloss = -2843.2, total_loss = -282.2 (0.69)\n",
            "epoch: 17943, dloss = 4.8, gloss = -2807.5, total_loss = -278.4 (0.70)\n",
            "epoch: 17944, dloss = 6.6, gloss = -2829.1, total_loss = -279.6 (0.69)\n",
            "epoch: 17945, dloss = 6.3, gloss = -2825.4, total_loss = -279.4 (0.67)\n",
            "epoch: 17946, dloss = 7.7, gloss = -2795.8, total_loss = -275.7 (0.69)\n",
            "epoch: 17947, dloss = 2.6, gloss = -2847.8, total_loss = -283.5 (0.67)\n",
            "epoch: 17948, dloss = 3.9, gloss = -2780.8, total_loss = -276.1 (0.65)\n",
            "epoch: 17949, dloss = 5.9, gloss = -2804.1, total_loss = -277.4 (0.77)\n",
            "epoch: 17950, dloss = -0.7, gloss = -2766.4, total_loss = -277.0 (0.92)\n",
            "epoch: 17951, dloss = 5.6, gloss = -2864.6, total_loss = -283.7 (0.69)\n",
            "epoch: 17952, dloss = 4.8, gloss = -2881.6, total_loss = -285.8 (0.68)\n",
            "epoch: 17953, dloss = 4.2, gloss = -2850.5, total_loss = -283.0 (0.68)\n",
            "epoch: 17954, dloss = 4.8, gloss = -2784.1, total_loss = -276.0 (0.68)\n",
            "epoch: 17955, dloss = 5.0, gloss = -2796.0, total_loss = -277.1 (0.68)\n",
            "epoch: 17956, dloss = 2.9, gloss = -2817.4, total_loss = -280.3 (0.68)\n",
            "epoch: 17957, dloss = 4.2, gloss = -2814.8, total_loss = -279.4 (0.69)\n",
            "epoch: 17958, dloss = 4.2, gloss = -2824.1, total_loss = -280.3 (0.69)\n",
            "epoch: 17959, dloss = 0.8, gloss = -2829.4, total_loss = -282.5 (0.71)\n",
            "epoch: 17960, dloss = 4.4, gloss = -2835.5, total_loss = -281.4 (0.68)\n",
            "epoch: 17961, dloss = -0.4, gloss = -2785.9, total_loss = -278.8 (0.67)\n",
            "epoch: 17962, dloss = 5.9, gloss = -2839.4, total_loss = -281.0 (0.70)\n",
            "epoch: 17963, dloss = 4.5, gloss = -2849.7, total_loss = -282.7 (0.71)\n",
            "epoch: 17964, dloss = 1.9, gloss = -2811.5, total_loss = -280.2 (0.68)\n",
            "epoch: 17965, dloss = 2.5, gloss = -2799.8, total_loss = -278.7 (0.93)\n",
            "epoch: 17966, dloss = 1.3, gloss = -2832.4, total_loss = -282.6 (0.80)\n",
            "epoch: 17967, dloss = 3.6, gloss = -2819.4, total_loss = -280.1 (0.66)\n",
            "epoch: 17968, dloss = 3.2, gloss = -2834.6, total_loss = -281.8 (0.68)\n",
            "epoch: 17969, dloss = 1.6, gloss = -2809.2, total_loss = -280.1 (0.69)\n",
            "epoch: 17970, dloss = 2.4, gloss = -2817.1, total_loss = -280.5 (0.68)\n",
            "epoch: 17971, dloss = 4.3, gloss = -2813.9, total_loss = -279.2 (0.67)\n",
            "epoch: 17972, dloss = 5.6, gloss = -2845.6, total_loss = -281.8 (0.70)\n",
            "epoch: 17973, dloss = 5.5, gloss = -2823.2, total_loss = -279.6 (0.68)\n",
            "epoch: 17974, dloss = 4.4, gloss = -2836.9, total_loss = -281.5 (0.66)\n",
            "epoch: 17975, dloss = 10.5, gloss = -2837.4, total_loss = -278.5 (0.66)\n",
            "epoch: 17976, dloss = 3.0, gloss = -2824.9, total_loss = -281.0 (0.68)\n",
            "epoch: 17977, dloss = 7.2, gloss = -2803.0, total_loss = -276.7 (0.68)\n",
            "epoch: 17978, dloss = 3.8, gloss = -2718.4, total_loss = -269.9 (0.69)\n",
            "epoch: 17979, dloss = 3.5, gloss = -2756.4, total_loss = -273.9 (0.69)\n",
            "epoch: 17980, dloss = 6.5, gloss = -2761.5, total_loss = -272.9 (0.70)\n",
            "epoch: 17981, dloss = 4.8, gloss = -2746.8, total_loss = -272.3 (1.04)\n",
            "epoch: 17982, dloss = 3.4, gloss = -2756.3, total_loss = -273.9 (0.69)\n",
            "epoch: 17983, dloss = 2.6, gloss = -2767.9, total_loss = -275.5 (0.70)\n",
            "epoch: 17984, dloss = 1.8, gloss = -2813.9, total_loss = -280.5 (0.68)\n",
            "epoch: 17985, dloss = 4.0, gloss = -2781.4, total_loss = -276.1 (0.67)\n",
            "epoch: 17986, dloss = 3.5, gloss = -2777.9, total_loss = -276.1 (0.68)\n",
            "epoch: 17987, dloss = 2.1, gloss = -2732.3, total_loss = -272.2 (0.69)\n",
            "epoch: 17988, dloss = 0.8, gloss = -2730.7, total_loss = -272.7 (0.67)\n",
            "epoch: 17989, dloss = 2.4, gloss = -2795.0, total_loss = -278.3 (0.67)\n",
            "epoch: 17990, dloss = 3.5, gloss = -2776.2, total_loss = -275.9 (0.68)\n",
            "epoch: 17991, dloss = 3.6, gloss = -2768.4, total_loss = -275.1 (0.68)\n",
            "epoch: 17992, dloss = 1.1, gloss = -2737.2, total_loss = -273.2 (0.70)\n",
            "epoch: 17993, dloss = 1.7, gloss = -2720.8, total_loss = -271.2 (0.68)\n",
            "epoch: 17994, dloss = 3.3, gloss = -2726.0, total_loss = -270.9 (0.68)\n",
            "epoch: 17995, dloss = 2.0, gloss = -2751.1, total_loss = -274.1 (0.69)\n",
            "epoch: 17996, dloss = 4.8, gloss = -2768.2, total_loss = -274.4 (0.79)\n",
            "epoch: 17997, dloss = 2.6, gloss = -2757.6, total_loss = -274.5 (0.93)\n",
            "epoch: 17998, dloss = -0.7, gloss = -2790.5, total_loss = -279.4 (0.69)\n",
            "epoch: 17999, dloss = 2.4, gloss = -2810.5, total_loss = -279.9 (0.70)\n",
            "epoch: 18000, dloss = 4.2, gloss = -2722.5, total_loss = -270.1 (0.69)\n",
            "epoch: 18001, dloss = 2.8, gloss = -2760.4, total_loss = -274.7 (0.70)\n",
            "epoch: 18002, dloss = 3.5, gloss = -2813.4, total_loss = -279.6 (0.69)\n",
            "epoch: 18003, dloss = 1.8, gloss = -2780.8, total_loss = -277.2 (0.70)\n",
            "epoch: 18004, dloss = 5.0, gloss = -2770.5, total_loss = -274.6 (0.67)\n",
            "epoch: 18005, dloss = 2.6, gloss = -2753.4, total_loss = -274.0 (0.68)\n",
            "epoch: 18006, dloss = 5.2, gloss = -2720.6, total_loss = -269.5 (0.68)\n",
            "epoch: 18007, dloss = 3.1, gloss = -2760.2, total_loss = -274.5 (0.69)\n",
            "epoch: 18008, dloss = 6.2, gloss = -2754.2, total_loss = -272.3 (0.69)\n",
            "epoch: 18009, dloss = 3.1, gloss = -2735.0, total_loss = -271.9 (0.69)\n",
            "epoch: 18010, dloss = 3.7, gloss = -2736.6, total_loss = -271.8 (0.66)\n",
            "epoch: 18011, dloss = 3.9, gloss = -2765.1, total_loss = -274.5 (0.68)\n",
            "epoch: 18012, dloss = 4.3, gloss = -2759.4, total_loss = -273.8 (0.91)\n",
            "epoch: 18013, dloss = 6.3, gloss = -2768.1, total_loss = -273.7 (0.82)\n",
            "epoch: 18014, dloss = 3.9, gloss = -2739.5, total_loss = -272.0 (0.69)\n",
            "epoch: 18015, dloss = 3.0, gloss = -2700.8, total_loss = -268.6 (0.69)\n",
            "epoch: 18016, dloss = 3.6, gloss = -2680.2, total_loss = -266.2 (0.69)\n",
            "epoch: 18017, dloss = 3.1, gloss = -2672.6, total_loss = -265.7 (0.68)\n",
            "epoch: 18018, dloss = 3.0, gloss = -2686.2, total_loss = -267.1 (0.70)\n",
            "epoch: 18019, dloss = -0.1, gloss = -2688.4, total_loss = -268.9 (0.68)\n",
            "epoch: 18020, dloss = 3.5, gloss = -2720.3, total_loss = -270.3 (0.68)\n",
            "epoch: 18021, dloss = -0.5, gloss = -2719.3, total_loss = -272.2 (0.67)\n",
            "epoch: 18022, dloss = 3.3, gloss = -2724.9, total_loss = -270.8 (0.70)\n",
            "epoch: 18023, dloss = 4.8, gloss = -2709.4, total_loss = -268.5 (0.68)\n",
            "epoch: 18024, dloss = 3.8, gloss = -2730.7, total_loss = -271.1 (0.67)\n",
            "epoch: 18025, dloss = 3.7, gloss = -2727.4, total_loss = -270.9 (0.67)\n",
            "epoch: 18026, dloss = 2.5, gloss = -2718.7, total_loss = -270.6 (0.69)\n",
            "epoch: 18027, dloss = 0.3, gloss = -2760.9, total_loss = -275.9 (0.75)\n",
            "epoch: 18028, dloss = 3.0, gloss = -2732.0, total_loss = -271.7 (1.00)\n",
            "epoch: 18029, dloss = 2.7, gloss = -2709.4, total_loss = -269.6 (0.67)\n",
            "epoch: 18030, dloss = 5.2, gloss = -2692.9, total_loss = -266.7 (0.67)\n",
            "epoch: 18031, dloss = 1.3, gloss = -2693.2, total_loss = -268.7 (0.67)\n",
            "epoch: 18032, dloss = 1.1, gloss = -2769.7, total_loss = -276.4 (0.68)\n",
            "epoch: 18033, dloss = 2.3, gloss = -2773.1, total_loss = -276.2 (0.70)\n",
            "epoch: 18034, dloss = 0.3, gloss = -2769.3, total_loss = -276.8 (0.68)\n",
            "epoch: 18035, dloss = 2.3, gloss = -2702.0, total_loss = -269.1 (0.68)\n",
            "epoch: 18036, dloss = 1.8, gloss = -2675.1, total_loss = -266.6 (0.68)\n",
            "epoch: 18037, dloss = 4.4, gloss = -2687.7, total_loss = -266.5 (0.71)\n",
            "epoch: 18038, dloss = 4.6, gloss = -2714.5, total_loss = -269.2 (0.69)\n",
            "epoch: 18039, dloss = 0.5, gloss = -2674.6, total_loss = -267.2 (0.66)\n",
            "epoch: 18040, dloss = 2.5, gloss = -2669.1, total_loss = -265.7 (0.70)\n",
            "epoch: 18041, dloss = 4.0, gloss = -2672.7, total_loss = -265.3 (0.69)\n",
            "epoch: 18042, dloss = -0.0, gloss = -2686.4, total_loss = -268.7 (0.68)\n",
            "epoch: 18043, dloss = 1.7, gloss = -2732.0, total_loss = -272.4 (0.91)\n",
            "epoch: 18044, dloss = 2.6, gloss = -2757.2, total_loss = -274.4 (0.83)\n",
            "epoch: 18045, dloss = 1.5, gloss = -2747.2, total_loss = -274.0 (0.67)\n",
            "epoch: 18046, dloss = 3.8, gloss = -2701.8, total_loss = -268.3 (0.69)\n",
            "epoch: 18047, dloss = 2.7, gloss = -2746.7, total_loss = -273.3 (0.69)\n",
            "epoch: 18048, dloss = 3.6, gloss = -2761.1, total_loss = -274.3 (0.69)\n",
            "epoch: 18049, dloss = 2.9, gloss = -2731.8, total_loss = -271.7 (0.68)\n",
            "epoch: 18050, dloss = 0.9, gloss = -2752.0, total_loss = -274.7 (0.68)\n",
            "epoch: 18051, dloss = 5.3, gloss = -2733.3, total_loss = -270.7 (0.69)\n",
            "epoch: 18052, dloss = 2.8, gloss = -2723.8, total_loss = -271.0 (0.68)\n",
            "epoch: 18053, dloss = 0.9, gloss = -2721.0, total_loss = -271.7 (0.68)\n",
            "epoch: 18054, dloss = 4.5, gloss = -2737.0, total_loss = -271.4 (0.68)\n",
            "epoch: 18055, dloss = 3.3, gloss = -2724.1, total_loss = -270.7 (0.67)\n",
            "epoch: 18056, dloss = 1.7, gloss = -2715.5, total_loss = -270.7 (0.68)\n",
            "epoch: 18057, dloss = 2.3, gloss = -2696.5, total_loss = -268.5 (0.69)\n",
            "epoch: 18058, dloss = 2.1, gloss = -2694.9, total_loss = -268.5 (0.69)\n",
            "epoch: 18059, dloss = 4.4, gloss = -2700.1, total_loss = -267.8 (1.01)\n",
            "epoch: 18060, dloss = -0.4, gloss = -2680.3, total_loss = -268.2 (0.71)\n",
            "epoch: 18061, dloss = -1.2, gloss = -2702.4, total_loss = -270.8 (0.69)\n",
            "epoch: 18062, dloss = 1.9, gloss = -2698.1, total_loss = -268.9 (0.68)\n",
            "epoch: 18063, dloss = -1.0, gloss = -2694.2, total_loss = -269.9 (0.67)\n",
            "epoch: 18064, dloss = -1.2, gloss = -2639.1, total_loss = -264.5 (0.67)\n",
            "epoch: 18065, dloss = 1.4, gloss = -2645.4, total_loss = -263.8 (0.69)\n",
            "epoch: 18066, dloss = 1.2, gloss = -2661.7, total_loss = -265.6 (0.67)\n",
            "epoch: 18067, dloss = 3.3, gloss = -2657.0, total_loss = -264.1 (0.68)\n",
            "epoch: 18068, dloss = 3.8, gloss = -2691.8, total_loss = -267.3 (0.67)\n",
            "epoch: 18069, dloss = 3.2, gloss = -2723.5, total_loss = -270.8 (0.69)\n",
            "epoch: 18070, dloss = 3.1, gloss = -2723.3, total_loss = -270.8 (0.66)\n",
            "epoch: 18071, dloss = 2.2, gloss = -2688.4, total_loss = -267.7 (1.26)\n",
            "epoch: 18072, dloss = 3.9, gloss = -2661.9, total_loss = -264.3 (0.70)\n",
            "epoch: 18073, dloss = 3.1, gloss = -2658.9, total_loss = -264.4 (0.72)\n",
            "epoch: 18074, dloss = 2.9, gloss = -2643.8, total_loss = -262.9 (1.00)\n",
            "epoch: 18075, dloss = 3.0, gloss = -2668.2, total_loss = -265.3 (0.71)\n",
            "epoch: 18076, dloss = 2.5, gloss = -2652.8, total_loss = -264.1 (0.69)\n",
            "epoch: 18077, dloss = 5.6, gloss = -2612.1, total_loss = -258.4 (0.68)\n",
            "epoch: 18078, dloss = 3.5, gloss = -2623.9, total_loss = -260.7 (0.67)\n",
            "epoch: 18079, dloss = 2.8, gloss = -2611.2, total_loss = -259.7 (0.68)\n",
            "epoch: 18080, dloss = 1.5, gloss = -2627.1, total_loss = -262.0 (0.67)\n",
            "epoch: 18081, dloss = 5.8, gloss = -2659.3, total_loss = -263.0 (0.69)\n",
            "epoch: 18082, dloss = 4.2, gloss = -2660.1, total_loss = -263.9 (0.70)\n",
            "epoch: 18083, dloss = 2.6, gloss = -2655.0, total_loss = -264.2 (0.67)\n",
            "epoch: 18084, dloss = 4.2, gloss = -2684.6, total_loss = -266.4 (0.69)\n",
            "epoch: 18085, dloss = 6.2, gloss = -2658.1, total_loss = -262.7 (0.69)\n",
            "epoch: 18086, dloss = 2.9, gloss = -2696.0, total_loss = -268.2 (0.68)\n",
            "epoch: 18087, dloss = 3.0, gloss = -2678.4, total_loss = -266.3 (0.67)\n",
            "epoch: 18088, dloss = 3.3, gloss = -2637.5, total_loss = -262.1 (0.65)\n",
            "epoch: 18089, dloss = 3.6, gloss = -2630.3, total_loss = -261.2 (0.85)\n",
            "epoch: 18090, dloss = 2.9, gloss = -2661.0, total_loss = -264.6 (0.87)\n",
            "epoch: 18091, dloss = 4.9, gloss = -2657.0, total_loss = -263.2 (0.68)\n",
            "epoch: 18092, dloss = 6.8, gloss = -2696.2, total_loss = -266.2 (0.86)\n",
            "epoch: 18093, dloss = 6.1, gloss = -2668.1, total_loss = -263.8 (0.89)\n",
            "epoch: 18094, dloss = 5.1, gloss = -2633.3, total_loss = -260.8 (0.67)\n",
            "epoch: 18095, dloss = 4.7, gloss = -2647.8, total_loss = -262.4 (0.67)\n",
            "epoch: 18096, dloss = 11.7, gloss = -2681.7, total_loss = -262.3 (0.68)\n",
            "epoch: 18097, dloss = 8.5, gloss = -2712.6, total_loss = -267.0 (0.67)\n",
            "epoch: 18098, dloss = 7.8, gloss = -2719.9, total_loss = -268.1 (0.69)\n",
            "epoch: 18099, dloss = 7.8, gloss = -2775.0, total_loss = -273.6 (0.91)\n",
            "epoch: 18100, dloss = 8.1, gloss = -2803.4, total_loss = -276.3 (0.84)\n",
            "epoch: 18101, dloss = 7.5, gloss = -2798.3, total_loss = -276.1 (0.66)\n",
            "epoch: 18102, dloss = 4.8, gloss = -2773.1, total_loss = -274.9 (0.67)\n",
            "epoch: 18103, dloss = 3.4, gloss = -2796.4, total_loss = -277.9 (0.70)\n",
            "epoch: 18104, dloss = 7.1, gloss = -2809.3, total_loss = -277.4 (0.96)\n",
            "epoch: 18105, dloss = 6.6, gloss = -2854.0, total_loss = -282.1 (0.78)\n",
            "epoch: 18106, dloss = 5.1, gloss = -2881.9, total_loss = -285.6 (0.68)\n",
            "epoch: 18107, dloss = 7.0, gloss = -2892.4, total_loss = -285.7 (0.69)\n",
            "epoch: 18108, dloss = 7.0, gloss = -2880.4, total_loss = -284.5 (0.71)\n",
            "epoch: 18109, dloss = 5.4, gloss = -2871.7, total_loss = -284.5 (0.68)\n",
            "epoch: 18110, dloss = 5.0, gloss = -2878.0, total_loss = -285.3 (0.68)\n",
            "epoch: 18111, dloss = 6.9, gloss = -2873.6, total_loss = -283.9 (0.69)\n",
            "epoch: 18112, dloss = 6.9, gloss = -2887.2, total_loss = -285.3 (0.69)\n",
            "epoch: 18113, dloss = 6.7, gloss = -2914.9, total_loss = -288.1 (0.69)\n",
            "epoch: 18114, dloss = 3.7, gloss = -2929.9, total_loss = -291.1 (0.67)\n",
            "epoch: 18115, dloss = 6.1, gloss = -2928.7, total_loss = -289.8 (0.67)\n",
            "epoch: 18116, dloss = 5.0, gloss = -2938.2, total_loss = -291.3 (0.69)\n",
            "epoch: 18117, dloss = 2.7, gloss = -2917.6, total_loss = -290.4 (0.69)\n",
            "epoch: 18118, dloss = 3.4, gloss = -2912.9, total_loss = -289.6 (0.70)\n",
            "epoch: 18119, dloss = 1.5, gloss = -2854.9, total_loss = -284.8 (0.76)\n",
            "epoch: 18120, dloss = 3.3, gloss = -2863.0, total_loss = -284.7 (0.98)\n",
            "epoch: 18121, dloss = 3.1, gloss = -2877.8, total_loss = -286.3 (0.70)\n",
            "epoch: 18122, dloss = -0.5, gloss = -2868.3, total_loss = -287.1 (0.67)\n",
            "epoch: 18123, dloss = 13.8, gloss = -2875.7, total_loss = -280.7 (0.68)\n",
            "epoch: 18124, dloss = 4.8, gloss = -2873.9, total_loss = -285.0 (0.71)\n",
            "epoch: 18125, dloss = 4.3, gloss = -2868.2, total_loss = -284.7 (0.68)\n",
            "epoch: 18126, dloss = 4.0, gloss = -2830.5, total_loss = -281.1 (0.69)\n",
            "epoch: 18127, dloss = 4.8, gloss = -2845.7, total_loss = -282.2 (0.70)\n",
            "epoch: 18128, dloss = 6.4, gloss = -2839.6, total_loss = -280.7 (0.68)\n",
            "epoch: 18129, dloss = 4.9, gloss = -2855.4, total_loss = -283.1 (0.72)\n",
            "epoch: 18130, dloss = 5.1, gloss = -2860.3, total_loss = -283.5 (0.69)\n",
            "epoch: 18131, dloss = 7.7, gloss = -2868.6, total_loss = -283.0 (0.68)\n",
            "epoch: 18132, dloss = 3.0, gloss = -2875.6, total_loss = -286.1 (0.72)\n",
            "epoch: 18133, dloss = 4.4, gloss = -2892.0, total_loss = -287.0 (0.68)\n",
            "epoch: 18134, dloss = 4.2, gloss = -2911.8, total_loss = -289.1 (0.69)\n",
            "epoch: 18135, dloss = 1.9, gloss = -2921.3, total_loss = -291.2 (0.96)\n",
            "epoch: 18136, dloss = 3.2, gloss = -2946.7, total_loss = -293.1 (0.78)\n",
            "epoch: 18137, dloss = 4.3, gloss = -2879.5, total_loss = -285.8 (0.69)\n",
            "epoch: 18138, dloss = 3.9, gloss = -2912.6, total_loss = -289.3 (0.70)\n",
            "epoch: 18139, dloss = 1.4, gloss = -2868.6, total_loss = -286.1 (0.67)\n",
            "epoch: 18140, dloss = 4.0, gloss = -2841.9, total_loss = -282.2 (0.69)\n",
            "epoch: 18141, dloss = 7.0, gloss = -2821.8, total_loss = -278.7 (0.69)\n",
            "epoch: 18142, dloss = 4.0, gloss = -2810.3, total_loss = -279.0 (0.67)\n",
            "epoch: 18143, dloss = 4.7, gloss = -2836.9, total_loss = -281.3 (0.69)\n",
            "epoch: 18144, dloss = 4.8, gloss = -2846.6, total_loss = -282.3 (0.71)\n",
            "epoch: 18145, dloss = 3.8, gloss = -2830.5, total_loss = -281.2 (0.66)\n",
            "epoch: 18146, dloss = 4.3, gloss = -2819.3, total_loss = -279.8 (0.68)\n",
            "epoch: 18147, dloss = 0.7, gloss = -2827.6, total_loss = -282.4 (0.68)\n",
            "epoch: 18148, dloss = 5.8, gloss = -2875.8, total_loss = -284.7 (0.68)\n",
            "epoch: 18149, dloss = 3.9, gloss = -2822.6, total_loss = -280.3 (0.68)\n",
            "epoch: 18150, dloss = 6.0, gloss = -2878.4, total_loss = -284.8 (0.76)\n",
            "epoch: 18151, dloss = 5.2, gloss = -2862.0, total_loss = -283.6 (0.98)\n",
            "epoch: 18152, dloss = 4.7, gloss = -2850.1, total_loss = -282.6 (0.76)\n",
            "epoch: 18153, dloss = 4.2, gloss = -2893.2, total_loss = -287.2 (1.06)\n",
            "epoch: 18154, dloss = 3.2, gloss = -2896.6, total_loss = -288.1 (0.69)\n",
            "epoch: 18155, dloss = 4.8, gloss = -2874.4, total_loss = -285.0 (0.71)\n",
            "epoch: 18156, dloss = 4.4, gloss = -2881.6, total_loss = -286.0 (0.67)\n",
            "epoch: 18157, dloss = 4.2, gloss = -2872.2, total_loss = -285.1 (0.70)\n",
            "epoch: 18158, dloss = 2.8, gloss = -2874.1, total_loss = -286.0 (0.70)\n",
            "epoch: 18159, dloss = 5.6, gloss = -2863.8, total_loss = -283.6 (0.66)\n",
            "epoch: 18160, dloss = 5.7, gloss = -2877.3, total_loss = -284.9 (0.70)\n",
            "epoch: 18161, dloss = 4.0, gloss = -2834.1, total_loss = -281.4 (0.69)\n",
            "epoch: 18162, dloss = 2.3, gloss = -2858.6, total_loss = -284.7 (0.69)\n",
            "epoch: 18163, dloss = 3.5, gloss = -2874.3, total_loss = -285.7 (0.68)\n",
            "epoch: 18164, dloss = 1.6, gloss = -2844.2, total_loss = -283.6 (0.69)\n",
            "epoch: 18165, dloss = 4.0, gloss = -2848.6, total_loss = -282.9 (1.26)\n",
            "epoch: 18166, dloss = 3.0, gloss = -2845.4, total_loss = -283.0 (0.69)\n",
            "epoch: 18167, dloss = 4.4, gloss = -2831.3, total_loss = -280.9 (0.67)\n",
            "epoch: 18168, dloss = 4.2, gloss = -2873.9, total_loss = -285.3 (0.68)\n",
            "epoch: 18169, dloss = 2.8, gloss = -2861.5, total_loss = -284.7 (0.68)\n",
            "epoch: 18170, dloss = 6.4, gloss = -2826.5, total_loss = -279.5 (0.68)\n",
            "epoch: 18171, dloss = 3.1, gloss = -2838.4, total_loss = -282.3 (0.70)\n",
            "epoch: 18172, dloss = 4.3, gloss = -2820.6, total_loss = -279.9 (0.69)\n",
            "epoch: 18173, dloss = 5.3, gloss = -2844.7, total_loss = -281.8 (0.69)\n",
            "epoch: 18174, dloss = 2.8, gloss = -2810.3, total_loss = -279.6 (0.71)\n",
            "epoch: 18175, dloss = 5.1, gloss = -2793.1, total_loss = -276.8 (0.70)\n",
            "epoch: 18176, dloss = 4.4, gloss = -2805.1, total_loss = -278.3 (0.70)\n",
            "epoch: 18177, dloss = 5.3, gloss = -2791.6, total_loss = -276.5 (0.67)\n",
            "epoch: 18178, dloss = 3.7, gloss = -2788.4, total_loss = -277.0 (0.70)\n",
            "epoch: 18179, dloss = 5.0, gloss = -2759.0, total_loss = -273.4 (0.67)\n",
            "epoch: 18180, dloss = 3.8, gloss = -2787.8, total_loss = -276.9 (0.87)\n",
            "epoch: 18181, dloss = 4.7, gloss = -2800.1, total_loss = -277.7 (0.86)\n",
            "epoch: 18182, dloss = 6.6, gloss = -2823.4, total_loss = -279.0 (0.69)\n",
            "epoch: 18183, dloss = 4.0, gloss = -2790.9, total_loss = -277.1 (0.68)\n",
            "epoch: 18184, dloss = 0.8, gloss = -2783.1, total_loss = -277.9 (0.68)\n",
            "epoch: 18185, dloss = 2.9, gloss = -2729.3, total_loss = -271.5 (0.68)\n",
            "epoch: 18186, dloss = 1.1, gloss = -2751.2, total_loss = -274.5 (0.69)\n",
            "epoch: 18187, dloss = 4.5, gloss = -2793.8, total_loss = -277.1 (0.70)\n",
            "epoch: 18188, dloss = 2.8, gloss = -2766.1, total_loss = -275.2 (0.69)\n",
            "epoch: 18189, dloss = 3.2, gloss = -2830.1, total_loss = -281.4 (0.68)\n",
            "epoch: 18190, dloss = 4.1, gloss = -2781.2, total_loss = -276.1 (0.70)\n",
            "epoch: 18191, dloss = 3.7, gloss = -2777.9, total_loss = -275.9 (0.70)\n",
            "epoch: 18192, dloss = 4.9, gloss = -2795.7, total_loss = -277.1 (0.71)\n",
            "epoch: 18193, dloss = 7.7, gloss = -2808.5, total_loss = -277.0 (0.69)\n",
            "epoch: 18194, dloss = 8.1, gloss = -2781.6, total_loss = -274.1 (0.70)\n",
            "epoch: 18195, dloss = 4.0, gloss = -2759.4, total_loss = -273.9 (0.79)\n",
            "epoch: 18196, dloss = 4.0, gloss = -2786.4, total_loss = -276.7 (1.00)\n",
            "epoch: 18197, dloss = 1.4, gloss = -2783.2, total_loss = -277.6 (0.65)\n",
            "epoch: 18198, dloss = 4.3, gloss = -2783.2, total_loss = -276.2 (0.69)\n",
            "epoch: 18199, dloss = 3.0, gloss = -2779.4, total_loss = -276.4 (0.69)\n",
            "epoch: 18200, dloss = 4.9, gloss = -2765.1, total_loss = -274.0 (0.67)\n",
            "epoch: 18201, dloss = 5.0, gloss = -2743.3, total_loss = -271.8 (0.68)\n",
            "epoch: 18202, dloss = 0.9, gloss = -2729.6, total_loss = -272.5 (0.69)\n",
            "epoch: 18203, dloss = 4.3, gloss = -2723.7, total_loss = -270.2 (0.70)\n",
            "epoch: 18204, dloss = 3.3, gloss = -2777.6, total_loss = -276.1 (0.70)\n",
            "epoch: 18205, dloss = 2.4, gloss = -2785.4, total_loss = -277.3 (0.67)\n",
            "epoch: 18206, dloss = 2.6, gloss = -2777.1, total_loss = -276.4 (0.68)\n",
            "epoch: 18207, dloss = 4.2, gloss = -2768.5, total_loss = -274.7 (0.68)\n",
            "epoch: 18208, dloss = 6.2, gloss = -2733.5, total_loss = -270.3 (0.68)\n",
            "epoch: 18209, dloss = 2.7, gloss = -2748.7, total_loss = -273.5 (0.69)\n",
            "epoch: 18210, dloss = 3.6, gloss = -2776.3, total_loss = -275.8 (0.68)\n",
            "epoch: 18211, dloss = 2.8, gloss = -2740.0, total_loss = -272.6 (0.87)\n",
            "epoch: 18212, dloss = 1.3, gloss = -2753.4, total_loss = -274.7 (0.89)\n",
            "epoch: 18213, dloss = 4.0, gloss = -2741.3, total_loss = -272.1 (0.71)\n",
            "epoch: 18214, dloss = 2.7, gloss = -2750.8, total_loss = -273.7 (0.67)\n",
            "epoch: 18215, dloss = 3.4, gloss = -2755.6, total_loss = -273.8 (0.69)\n",
            "epoch: 18216, dloss = 5.2, gloss = -2804.6, total_loss = -277.9 (0.71)\n",
            "epoch: 18217, dloss = 4.8, gloss = -2766.9, total_loss = -274.3 (0.69)\n",
            "epoch: 18218, dloss = 4.9, gloss = -2778.1, total_loss = -275.4 (0.68)\n",
            "epoch: 18219, dloss = 2.7, gloss = -2835.3, total_loss = -282.2 (0.69)\n",
            "epoch: 18220, dloss = 5.7, gloss = -2792.1, total_loss = -276.3 (0.68)\n",
            "epoch: 18221, dloss = 5.2, gloss = -2801.9, total_loss = -277.6 (0.69)\n",
            "epoch: 18222, dloss = 3.1, gloss = -2765.5, total_loss = -275.0 (0.67)\n",
            "epoch: 18223, dloss = 1.8, gloss = -2759.6, total_loss = -275.1 (0.68)\n",
            "epoch: 18224, dloss = 2.6, gloss = -2774.8, total_loss = -276.2 (0.67)\n",
            "epoch: 18225, dloss = 4.2, gloss = -2799.9, total_loss = -277.9 (0.69)\n",
            "epoch: 18226, dloss = 5.0, gloss = -2780.6, total_loss = -275.6 (0.68)\n",
            "epoch: 18227, dloss = 2.3, gloss = -2760.2, total_loss = -274.9 (1.04)\n",
            "epoch: 18228, dloss = 3.0, gloss = -2766.8, total_loss = -275.2 (0.68)\n",
            "epoch: 18229, dloss = 3.5, gloss = -2794.1, total_loss = -277.7 (0.69)\n",
            "epoch: 18230, dloss = 4.3, gloss = -2750.8, total_loss = -273.0 (0.68)\n",
            "epoch: 18231, dloss = 5.3, gloss = -2763.7, total_loss = -273.7 (0.68)\n",
            "epoch: 18232, dloss = 4.2, gloss = -2740.9, total_loss = -272.0 (0.70)\n",
            "epoch: 18233, dloss = 2.8, gloss = -2754.8, total_loss = -274.1 (0.69)\n",
            "epoch: 18234, dloss = 3.3, gloss = -2762.7, total_loss = -274.6 (0.68)\n",
            "epoch: 18235, dloss = 4.4, gloss = -2787.6, total_loss = -276.6 (0.69)\n",
            "epoch: 18236, dloss = 4.4, gloss = -2753.8, total_loss = -273.2 (0.67)\n",
            "epoch: 18237, dloss = 5.6, gloss = -2764.0, total_loss = -273.6 (0.68)\n",
            "epoch: 18238, dloss = 3.8, gloss = -2756.8, total_loss = -273.8 (0.69)\n",
            "epoch: 18239, dloss = 6.5, gloss = -2762.8, total_loss = -273.0 (0.67)\n",
            "epoch: 18240, dloss = 4.0, gloss = -2759.9, total_loss = -274.0 (0.69)\n",
            "epoch: 18241, dloss = 3.9, gloss = -2755.2, total_loss = -273.6 (0.67)\n",
            "epoch: 18242, dloss = 2.0, gloss = -2737.2, total_loss = -272.7 (0.82)\n",
            "epoch: 18243, dloss = -0.2, gloss = -2769.6, total_loss = -277.1 (0.91)\n",
            "epoch: 18244, dloss = 1.6, gloss = -2786.0, total_loss = -277.8 (0.67)\n",
            "epoch: 18245, dloss = 3.5, gloss = -2828.8, total_loss = -281.1 (0.69)\n",
            "epoch: 18246, dloss = 6.5, gloss = -2781.8, total_loss = -274.9 (0.70)\n",
            "epoch: 18247, dloss = 5.2, gloss = -2750.8, total_loss = -272.5 (0.68)\n",
            "epoch: 18248, dloss = 2.4, gloss = -2761.0, total_loss = -274.9 (0.70)\n",
            "epoch: 18249, dloss = 3.2, gloss = -2793.9, total_loss = -277.8 (0.67)\n",
            "epoch: 18250, dloss = 3.4, gloss = -2783.6, total_loss = -276.7 (0.68)\n",
            "epoch: 18251, dloss = 3.6, gloss = -2784.0, total_loss = -276.6 (0.68)\n",
            "epoch: 18252, dloss = 0.4, gloss = -2813.4, total_loss = -281.1 (0.67)\n",
            "epoch: 18253, dloss = 2.9, gloss = -2776.1, total_loss = -276.2 (0.69)\n",
            "epoch: 18254, dloss = 3.8, gloss = -2753.0, total_loss = -273.4 (0.68)\n",
            "epoch: 18255, dloss = 4.1, gloss = -2748.6, total_loss = -272.8 (0.68)\n",
            "epoch: 18256, dloss = 5.9, gloss = -2728.9, total_loss = -269.9 (0.68)\n",
            "epoch: 18257, dloss = 0.6, gloss = -2685.9, total_loss = -268.3 (0.68)\n",
            "epoch: 18258, dloss = 3.1, gloss = -2691.9, total_loss = -267.7 (0.97)\n",
            "epoch: 18259, dloss = 2.5, gloss = -2703.2, total_loss = -269.1 (0.77)\n",
            "epoch: 18260, dloss = 5.3, gloss = -2726.9, total_loss = -270.0 (0.70)\n",
            "epoch: 18261, dloss = 4.1, gloss = -2778.0, total_loss = -275.8 (0.69)\n",
            "epoch: 18262, dloss = 4.3, gloss = -2757.9, total_loss = -273.6 (0.68)\n",
            "epoch: 18263, dloss = 5.4, gloss = -2731.3, total_loss = -270.4 (0.67)\n",
            "epoch: 18264, dloss = 2.6, gloss = -2717.5, total_loss = -270.5 (0.70)\n",
            "epoch: 18265, dloss = 0.9, gloss = -2725.7, total_loss = -272.1 (0.67)\n",
            "epoch: 18266, dloss = 1.9, gloss = -2723.3, total_loss = -271.4 (0.68)\n",
            "epoch: 18267, dloss = 1.7, gloss = -2699.2, total_loss = -269.1 (0.69)\n",
            "epoch: 18268, dloss = 3.9, gloss = -2725.0, total_loss = -270.6 (0.67)\n",
            "epoch: 18269, dloss = 3.6, gloss = -2726.5, total_loss = -270.8 (0.71)\n",
            "epoch: 18270, dloss = 3.3, gloss = -2695.4, total_loss = -267.9 (0.68)\n",
            "epoch: 18271, dloss = 3.0, gloss = -2687.4, total_loss = -267.2 (0.67)\n",
            "epoch: 18272, dloss = 4.7, gloss = -2660.1, total_loss = -263.7 (0.67)\n",
            "epoch: 18273, dloss = 6.4, gloss = -2693.0, total_loss = -266.1 (1.26)\n",
            "epoch: 18274, dloss = 5.0, gloss = -2691.0, total_loss = -266.6 (0.70)\n",
            "epoch: 18275, dloss = 3.5, gloss = -2714.6, total_loss = -269.7 (0.68)\n",
            "epoch: 18276, dloss = 4.4, gloss = -2746.2, total_loss = -272.4 (0.70)\n",
            "epoch: 18277, dloss = 3.1, gloss = -2734.0, total_loss = -271.8 (0.70)\n",
            "epoch: 18278, dloss = 2.5, gloss = -2702.0, total_loss = -268.9 (0.68)\n",
            "epoch: 18279, dloss = 4.4, gloss = -2728.3, total_loss = -270.6 (0.69)\n",
            "epoch: 18280, dloss = 3.2, gloss = -2653.0, total_loss = -263.7 (0.69)\n",
            "epoch: 18281, dloss = 4.6, gloss = -2657.5, total_loss = -263.4 (0.68)\n",
            "epoch: 18282, dloss = 1.5, gloss = -2660.5, total_loss = -265.3 (0.69)\n",
            "epoch: 18283, dloss = 3.6, gloss = -2649.1, total_loss = -263.1 (0.70)\n",
            "epoch: 18284, dloss = 3.8, gloss = -2648.2, total_loss = -262.9 (0.71)\n",
            "epoch: 18285, dloss = 1.8, gloss = -2648.1, total_loss = -263.9 (0.70)\n",
            "epoch: 18286, dloss = 0.7, gloss = -2652.5, total_loss = -264.9 (0.67)\n",
            "epoch: 18287, dloss = 1.3, gloss = -2642.9, total_loss = -263.6 (0.68)\n",
            "epoch: 18288, dloss = 1.3, gloss = -2654.1, total_loss = -264.8 (0.90)\n",
            "epoch: 18289, dloss = 5.7, gloss = -2662.3, total_loss = -263.4 (0.83)\n",
            "epoch: 18290, dloss = 3.6, gloss = -2659.9, total_loss = -264.2 (0.67)\n",
            "epoch: 18291, dloss = 4.8, gloss = -2627.2, total_loss = -260.3 (0.66)\n",
            "epoch: 18292, dloss = 0.9, gloss = -2673.9, total_loss = -266.9 (0.68)\n",
            "epoch: 18293, dloss = 3.3, gloss = -2707.4, total_loss = -269.1 (0.68)\n",
            "epoch: 18294, dloss = 5.3, gloss = -2693.0, total_loss = -266.7 (0.70)\n",
            "epoch: 18295, dloss = 6.0, gloss = -2709.5, total_loss = -267.9 (0.67)\n",
            "epoch: 18296, dloss = 6.2, gloss = -2765.8, total_loss = -273.5 (0.71)\n",
            "epoch: 18297, dloss = 7.3, gloss = -2722.0, total_loss = -268.6 (0.69)\n",
            "epoch: 18298, dloss = 5.9, gloss = -2758.6, total_loss = -272.9 (0.69)\n",
            "epoch: 18299, dloss = 0.7, gloss = -2761.8, total_loss = -275.8 (0.70)\n",
            "epoch: 18300, dloss = 5.3, gloss = -2784.2, total_loss = -275.8 (0.69)\n",
            "epoch: 18301, dloss = 5.0, gloss = -2755.7, total_loss = -273.1 (0.68)\n",
            "epoch: 18302, dloss = 5.9, gloss = -2767.4, total_loss = -273.8 (0.69)\n",
            "epoch: 18303, dloss = 5.6, gloss = -2770.0, total_loss = -274.2 (0.71)\n",
            "epoch: 18304, dloss = 3.8, gloss = -2745.8, total_loss = -272.7 (1.03)\n",
            "epoch: 18305, dloss = 3.1, gloss = -2736.7, total_loss = -272.1 (0.68)\n",
            "epoch: 18306, dloss = 1.1, gloss = -2692.5, total_loss = -268.7 (0.66)\n",
            "epoch: 18307, dloss = 5.1, gloss = -2697.7, total_loss = -267.2 (0.67)\n",
            "epoch: 18308, dloss = 5.4, gloss = -2727.9, total_loss = -270.1 (0.68)\n",
            "epoch: 18309, dloss = 4.3, gloss = -2694.1, total_loss = -267.3 (0.89)\n",
            "epoch: 18310, dloss = 1.7, gloss = -2717.2, total_loss = -270.9 (1.26)\n",
            "epoch: 18311, dloss = 3.6, gloss = -2723.0, total_loss = -270.5 (1.11)\n",
            "epoch: 18312, dloss = 4.7, gloss = -2720.9, total_loss = -269.7 (0.74)\n",
            "epoch: 18313, dloss = 5.3, gloss = -2722.2, total_loss = -269.6 (0.65)\n",
            "epoch: 18314, dloss = 4.8, gloss = -2691.4, total_loss = -266.8 (0.69)\n",
            "epoch: 18315, dloss = 4.8, gloss = -2681.2, total_loss = -265.7 (0.67)\n",
            "epoch: 18316, dloss = 5.0, gloss = -2681.6, total_loss = -265.7 (0.67)\n",
            "epoch: 18317, dloss = 0.7, gloss = -2687.6, total_loss = -268.4 (0.76)\n",
            "epoch: 18318, dloss = 2.6, gloss = -2679.1, total_loss = -266.6 (0.98)\n",
            "epoch: 18319, dloss = 1.9, gloss = -2691.8, total_loss = -268.2 (0.69)\n",
            "epoch: 18320, dloss = 4.0, gloss = -2715.1, total_loss = -269.5 (0.67)\n",
            "epoch: 18321, dloss = 3.5, gloss = -2749.9, total_loss = -273.2 (0.69)\n",
            "epoch: 18322, dloss = 3.9, gloss = -2740.3, total_loss = -272.1 (0.69)\n",
            "epoch: 18323, dloss = 4.5, gloss = -2749.8, total_loss = -272.7 (0.67)\n",
            "epoch: 18324, dloss = 2.1, gloss = -2770.2, total_loss = -276.0 (0.69)\n",
            "epoch: 18325, dloss = 1.3, gloss = -2732.8, total_loss = -272.6 (0.66)\n",
            "epoch: 18326, dloss = 2.0, gloss = -2726.2, total_loss = -271.6 (0.70)\n",
            "epoch: 18327, dloss = 2.7, gloss = -2692.8, total_loss = -267.9 (0.69)\n",
            "epoch: 18328, dloss = 3.6, gloss = -2720.4, total_loss = -270.3 (0.69)\n",
            "epoch: 18329, dloss = 4.6, gloss = -2744.2, total_loss = -272.1 (0.70)\n",
            "epoch: 18330, dloss = 6.0, gloss = -2724.3, total_loss = -269.4 (0.68)\n",
            "epoch: 18331, dloss = 3.9, gloss = -2762.6, total_loss = -274.3 (0.70)\n",
            "epoch: 18332, dloss = 3.9, gloss = -2769.1, total_loss = -275.0 (0.69)\n",
            "epoch: 18333, dloss = 5.8, gloss = -2733.8, total_loss = -270.5 (0.92)\n",
            "epoch: 18334, dloss = 5.1, gloss = -2733.3, total_loss = -270.8 (1.27)\n",
            "epoch: 18335, dloss = 3.3, gloss = -2732.8, total_loss = -271.6 (0.70)\n",
            "epoch: 18336, dloss = 5.4, gloss = -2740.6, total_loss = -271.4 (0.69)\n",
            "epoch: 18337, dloss = 2.7, gloss = -2755.3, total_loss = -274.2 (0.70)\n",
            "epoch: 18338, dloss = 5.9, gloss = -2745.7, total_loss = -271.6 (0.73)\n",
            "epoch: 18339, dloss = 0.5, gloss = -2756.1, total_loss = -275.4 (1.05)\n",
            "epoch: 18340, dloss = 5.5, gloss = -2757.1, total_loss = -273.0 (0.68)\n",
            "epoch: 18341, dloss = 0.6, gloss = -2704.3, total_loss = -270.1 (0.70)\n",
            "epoch: 18342, dloss = 2.9, gloss = -2743.2, total_loss = -272.9 (0.70)\n",
            "epoch: 18343, dloss = 2.7, gloss = -2744.4, total_loss = -273.1 (0.69)\n",
            "epoch: 18344, dloss = 3.8, gloss = -2702.2, total_loss = -268.3 (0.69)\n",
            "epoch: 18345, dloss = 4.0, gloss = -2705.1, total_loss = -268.5 (0.67)\n",
            "epoch: 18346, dloss = 6.0, gloss = -2758.4, total_loss = -272.9 (0.69)\n",
            "epoch: 18347, dloss = 3.5, gloss = -2737.0, total_loss = -271.9 (0.84)\n",
            "epoch: 18348, dloss = 4.7, gloss = -2729.2, total_loss = -270.6 (1.10)\n",
            "epoch: 18349, dloss = 5.0, gloss = -2730.0, total_loss = -270.5 (0.84)\n",
            "epoch: 18350, dloss = 2.1, gloss = -2742.0, total_loss = -273.2 (0.68)\n",
            "epoch: 18351, dloss = 2.4, gloss = -2733.8, total_loss = -272.2 (0.70)\n",
            "epoch: 18352, dloss = 2.4, gloss = -2733.6, total_loss = -272.1 (0.69)\n",
            "epoch: 18353, dloss = 2.4, gloss = -2686.4, total_loss = -267.5 (0.69)\n",
            "epoch: 18354, dloss = 4.5, gloss = -2754.7, total_loss = -273.2 (0.70)\n",
            "epoch: 18355, dloss = 4.4, gloss = -2740.3, total_loss = -271.9 (0.68)\n",
            "epoch: 18356, dloss = 6.6, gloss = -2748.4, total_loss = -271.6 (0.69)\n",
            "epoch: 18357, dloss = 1.7, gloss = -2702.0, total_loss = -269.4 (0.67)\n",
            "epoch: 18358, dloss = 4.9, gloss = -2697.5, total_loss = -267.3 (0.67)\n",
            "epoch: 18359, dloss = 3.3, gloss = -2717.3, total_loss = -270.1 (0.70)\n",
            "epoch: 18360, dloss = 1.3, gloss = -2708.9, total_loss = -270.3 (0.70)\n",
            "epoch: 18361, dloss = 1.3, gloss = -2741.0, total_loss = -273.4 (0.69)\n",
            "epoch: 18362, dloss = 3.3, gloss = -2740.6, total_loss = -272.4 (0.68)\n",
            "epoch: 18363, dloss = 4.0, gloss = -2701.3, total_loss = -268.1 (0.98)\n",
            "epoch: 18364, dloss = 5.1, gloss = -2702.7, total_loss = -267.7 (0.77)\n",
            "epoch: 18365, dloss = 2.9, gloss = -2688.3, total_loss = -267.4 (0.71)\n",
            "epoch: 18366, dloss = 2.9, gloss = -2716.2, total_loss = -270.2 (0.69)\n",
            "epoch: 18367, dloss = 3.3, gloss = -2712.3, total_loss = -269.6 (0.66)\n",
            "epoch: 18368, dloss = 4.6, gloss = -2732.1, total_loss = -270.9 (0.70)\n",
            "epoch: 18369, dloss = 2.7, gloss = -2746.1, total_loss = -273.3 (0.68)\n",
            "epoch: 18370, dloss = 6.2, gloss = -2775.1, total_loss = -274.4 (0.66)\n",
            "epoch: 18371, dloss = 4.6, gloss = -2766.8, total_loss = -274.4 (0.69)\n",
            "epoch: 18372, dloss = 5.6, gloss = -2757.9, total_loss = -273.0 (0.69)\n",
            "epoch: 18373, dloss = 5.1, gloss = -2745.2, total_loss = -271.9 (0.68)\n",
            "epoch: 18374, dloss = 4.1, gloss = -2721.3, total_loss = -270.1 (0.67)\n",
            "epoch: 18375, dloss = 3.7, gloss = -2720.8, total_loss = -270.2 (0.67)\n",
            "epoch: 18376, dloss = 3.5, gloss = -2733.5, total_loss = -271.6 (0.69)\n",
            "epoch: 18377, dloss = 3.2, gloss = -2746.9, total_loss = -273.1 (0.69)\n",
            "epoch: 18378, dloss = 4.5, gloss = -2721.6, total_loss = -269.9 (0.76)\n",
            "epoch: 18379, dloss = 4.0, gloss = -2719.1, total_loss = -269.9 (0.97)\n",
            "epoch: 18380, dloss = 3.2, gloss = -2745.2, total_loss = -272.9 (0.67)\n",
            "epoch: 18381, dloss = 4.5, gloss = -2751.4, total_loss = -272.9 (0.69)\n",
            "epoch: 18382, dloss = 3.9, gloss = -2728.1, total_loss = -270.9 (0.69)\n",
            "epoch: 18383, dloss = 7.0, gloss = -2729.4, total_loss = -269.4 (0.69)\n",
            "epoch: 18384, dloss = 0.9, gloss = -2729.5, total_loss = -272.5 (0.67)\n",
            "epoch: 18385, dloss = 7.7, gloss = -2760.2, total_loss = -272.2 (0.69)\n",
            "epoch: 18386, dloss = 4.2, gloss = -2742.1, total_loss = -272.1 (0.68)\n",
            "epoch: 18387, dloss = 5.0, gloss = -2730.3, total_loss = -270.5 (0.70)\n",
            "epoch: 18388, dloss = 7.0, gloss = -2712.5, total_loss = -267.7 (0.69)\n",
            "epoch: 18389, dloss = 3.4, gloss = -2725.5, total_loss = -270.8 (0.66)\n",
            "epoch: 18390, dloss = 6.1, gloss = -2743.4, total_loss = -271.3 (0.68)\n",
            "epoch: 18391, dloss = 10.0, gloss = -2706.3, total_loss = -265.6 (0.68)\n",
            "epoch: 18392, dloss = 1.9, gloss = -2744.9, total_loss = -273.5 (0.67)\n",
            "epoch: 18393, dloss = 4.3, gloss = -2758.7, total_loss = -273.7 (0.70)\n",
            "epoch: 18394, dloss = 5.4, gloss = -2765.6, total_loss = -273.9 (0.88)\n",
            "epoch: 18395, dloss = 6.7, gloss = -2740.2, total_loss = -270.7 (0.83)\n",
            "epoch: 18396, dloss = 5.2, gloss = -2776.3, total_loss = -275.0 (0.67)\n",
            "epoch: 18397, dloss = 4.5, gloss = -2768.0, total_loss = -274.6 (0.69)\n",
            "epoch: 18398, dloss = 9.2, gloss = -2800.7, total_loss = -275.5 (0.70)\n",
            "epoch: 18399, dloss = 10.6, gloss = -2737.3, total_loss = -268.4 (0.69)\n",
            "epoch: 18400, dloss = 4.8, gloss = -2740.9, total_loss = -271.7 (0.66)\n",
            "epoch: 18401, dloss = 7.1, gloss = -2766.0, total_loss = -273.0 (0.68)\n",
            "epoch: 18402, dloss = 5.0, gloss = -2767.8, total_loss = -274.3 (0.68)\n",
            "epoch: 18403, dloss = 5.0, gloss = -2783.1, total_loss = -275.8 (0.70)\n",
            "epoch: 18404, dloss = 14.2, gloss = -2754.1, total_loss = -268.3 (0.69)\n",
            "epoch: 18405, dloss = 8.9, gloss = -2775.5, total_loss = -273.1 (0.70)\n",
            "epoch: 18406, dloss = 4.8, gloss = -2745.4, total_loss = -272.1 (0.70)\n",
            "epoch: 18407, dloss = 5.8, gloss = -2761.9, total_loss = -273.3 (0.67)\n",
            "epoch: 18408, dloss = 5.0, gloss = -2756.7, total_loss = -273.2 (0.68)\n",
            "epoch: 18409, dloss = 4.8, gloss = -2785.1, total_loss = -276.1 (0.69)\n",
            "epoch: 18410, dloss = 6.2, gloss = -2795.9, total_loss = -276.5 (1.01)\n",
            "epoch: 18411, dloss = 10.3, gloss = -2743.5, total_loss = -269.2 (0.68)\n",
            "epoch: 18412, dloss = 4.0, gloss = -2755.9, total_loss = -273.6 (0.68)\n",
            "epoch: 18413, dloss = 3.8, gloss = -2774.6, total_loss = -275.6 (0.69)\n",
            "epoch: 18414, dloss = 3.5, gloss = -2780.3, total_loss = -276.3 (0.69)\n",
            "epoch: 18415, dloss = 5.1, gloss = -2753.1, total_loss = -272.7 (0.68)\n",
            "epoch: 18416, dloss = 5.0, gloss = -2766.4, total_loss = -274.2 (0.70)\n",
            "epoch: 18417, dloss = 6.6, gloss = -2779.4, total_loss = -274.7 (0.69)\n",
            "epoch: 18418, dloss = 5.5, gloss = -2748.4, total_loss = -272.1 (0.66)\n",
            "epoch: 18419, dloss = 6.0, gloss = -2729.8, total_loss = -270.0 (0.71)\n",
            "epoch: 18420, dloss = 13.2, gloss = -2726.6, total_loss = -266.1 (0.68)\n",
            "epoch: 18421, dloss = 5.0, gloss = -2731.7, total_loss = -270.7 (0.69)\n",
            "epoch: 18422, dloss = 2.4, gloss = -2731.1, total_loss = -271.9 (0.66)\n",
            "epoch: 18423, dloss = 5.3, gloss = -2717.1, total_loss = -269.0 (0.71)\n",
            "epoch: 18424, dloss = 3.9, gloss = -2729.4, total_loss = -271.0 (0.69)\n",
            "epoch: 18425, dloss = 5.8, gloss = -2745.4, total_loss = -271.7 (0.80)\n",
            "epoch: 18426, dloss = 3.1, gloss = -2727.6, total_loss = -271.2 (0.91)\n",
            "epoch: 18427, dloss = 3.2, gloss = -2742.2, total_loss = -272.6 (0.68)\n",
            "epoch: 18428, dloss = 3.8, gloss = -2765.7, total_loss = -274.7 (0.68)\n",
            "epoch: 18429, dloss = 5.4, gloss = -2722.1, total_loss = -269.5 (0.70)\n",
            "epoch: 18430, dloss = 4.1, gloss = -2762.9, total_loss = -274.2 (0.70)\n",
            "epoch: 18431, dloss = 6.8, gloss = -2763.5, total_loss = -273.0 (0.68)\n",
            "epoch: 18432, dloss = 5.7, gloss = -2732.9, total_loss = -270.4 (0.67)\n",
            "epoch: 18433, dloss = 6.5, gloss = -2727.1, total_loss = -269.5 (0.69)\n",
            "epoch: 18434, dloss = 7.6, gloss = -2743.5, total_loss = -270.5 (0.66)\n",
            "epoch: 18435, dloss = 3.9, gloss = -2741.0, total_loss = -272.2 (0.69)\n",
            "epoch: 18436, dloss = 10.7, gloss = -2720.7, total_loss = -266.7 (0.68)\n",
            "epoch: 18437, dloss = 4.2, gloss = -2749.1, total_loss = -272.8 (0.68)\n",
            "epoch: 18438, dloss = 5.7, gloss = -2719.7, total_loss = -269.1 (0.70)\n",
            "epoch: 18439, dloss = 7.0, gloss = -2765.8, total_loss = -273.1 (0.68)\n",
            "epoch: 18440, dloss = 4.9, gloss = -2794.5, total_loss = -277.0 (0.86)\n",
            "epoch: 18441, dloss = 6.2, gloss = -2803.8, total_loss = -277.3 (1.26)\n",
            "epoch: 18442, dloss = 5.1, gloss = -2803.6, total_loss = -277.8 (0.82)\n",
            "epoch: 18443, dloss = 3.2, gloss = -2741.9, total_loss = -272.6 (0.69)\n",
            "epoch: 18444, dloss = 4.4, gloss = -2780.7, total_loss = -275.9 (0.67)\n",
            "epoch: 18445, dloss = 5.4, gloss = -2811.9, total_loss = -278.5 (0.67)\n",
            "epoch: 18446, dloss = 4.7, gloss = -2782.9, total_loss = -275.9 (0.68)\n",
            "epoch: 18447, dloss = 5.4, gloss = -2809.7, total_loss = -278.3 (0.68)\n",
            "epoch: 18448, dloss = 4.3, gloss = -2830.3, total_loss = -280.9 (0.70)\n",
            "epoch: 18449, dloss = 6.3, gloss = -2834.2, total_loss = -280.3 (0.68)\n",
            "epoch: 18450, dloss = 3.3, gloss = -2806.3, total_loss = -279.0 (0.68)\n",
            "epoch: 18451, dloss = 8.8, gloss = -2758.0, total_loss = -271.4 (0.67)\n",
            "epoch: 18452, dloss = 5.4, gloss = -2761.9, total_loss = -273.5 (0.68)\n",
            "epoch: 18453, dloss = 4.8, gloss = -2760.5, total_loss = -273.6 (0.68)\n",
            "epoch: 18454, dloss = 5.1, gloss = -2779.9, total_loss = -275.4 (0.70)\n",
            "epoch: 18455, dloss = 6.3, gloss = -2812.7, total_loss = -278.1 (0.68)\n",
            "epoch: 18456, dloss = 5.7, gloss = -2820.3, total_loss = -279.2 (0.73)\n",
            "epoch: 18457, dloss = 4.6, gloss = -2830.8, total_loss = -280.8 (1.06)\n",
            "epoch: 18458, dloss = 4.6, gloss = -2852.8, total_loss = -283.0 (0.70)\n",
            "epoch: 18459, dloss = 3.6, gloss = -2899.8, total_loss = -288.2 (0.68)\n",
            "epoch: 18460, dloss = 4.2, gloss = -2897.4, total_loss = -287.6 (0.70)\n",
            "epoch: 18461, dloss = 4.6, gloss = -2893.9, total_loss = -287.1 (0.69)\n",
            "epoch: 18462, dloss = 3.9, gloss = -2880.8, total_loss = -286.1 (0.66)\n",
            "epoch: 18463, dloss = 4.4, gloss = -2907.7, total_loss = -288.6 (0.68)\n",
            "epoch: 18464, dloss = 4.9, gloss = -2923.2, total_loss = -289.9 (0.68)\n",
            "epoch: 18465, dloss = 3.8, gloss = -2858.6, total_loss = -284.0 (0.69)\n",
            "epoch: 18466, dloss = 6.0, gloss = -2858.4, total_loss = -282.9 (0.70)\n",
            "epoch: 18467, dloss = 6.5, gloss = -2840.6, total_loss = -280.8 (0.69)\n",
            "epoch: 18468, dloss = 5.2, gloss = -2804.2, total_loss = -277.8 (0.70)\n",
            "epoch: 18469, dloss = 4.9, gloss = -2847.4, total_loss = -282.3 (0.70)\n",
            "epoch: 18470, dloss = 4.1, gloss = -2829.3, total_loss = -280.9 (0.69)\n",
            "epoch: 18471, dloss = 4.1, gloss = -2867.8, total_loss = -284.7 (0.69)\n",
            "epoch: 18472, dloss = 4.2, gloss = -2854.9, total_loss = -283.4 (0.89)\n",
            "epoch: 18473, dloss = 2.5, gloss = -2845.3, total_loss = -283.3 (0.82)\n",
            "epoch: 18474, dloss = 4.7, gloss = -2822.3, total_loss = -279.9 (0.66)\n",
            "epoch: 18475, dloss = 5.2, gloss = -2842.1, total_loss = -281.6 (0.67)\n",
            "epoch: 18476, dloss = 1.6, gloss = -2826.3, total_loss = -281.8 (0.67)\n",
            "epoch: 18477, dloss = 1.1, gloss = -2800.7, total_loss = -279.5 (0.70)\n",
            "epoch: 18478, dloss = 6.0, gloss = -2818.9, total_loss = -278.9 (0.70)\n",
            "epoch: 18479, dloss = 4.8, gloss = -2843.5, total_loss = -281.9 (0.68)\n",
            "epoch: 18480, dloss = 15.7, gloss = -2829.8, total_loss = -275.1 (0.68)\n",
            "epoch: 18481, dloss = 6.3, gloss = -2830.5, total_loss = -279.9 (0.67)\n",
            "epoch: 18482, dloss = 5.6, gloss = -2847.4, total_loss = -281.9 (0.69)\n",
            "epoch: 18483, dloss = 3.2, gloss = -2863.9, total_loss = -284.8 (0.73)\n",
            "epoch: 18484, dloss = 4.7, gloss = -2819.6, total_loss = -279.6 (0.72)\n",
            "epoch: 18485, dloss = 3.8, gloss = -2814.8, total_loss = -279.6 (0.69)\n",
            "epoch: 18486, dloss = 4.4, gloss = -2801.9, total_loss = -278.0 (0.68)\n",
            "epoch: 18487, dloss = 5.5, gloss = -2776.8, total_loss = -274.9 (0.73)\n",
            "epoch: 18488, dloss = 2.1, gloss = -2776.2, total_loss = -276.6 (1.02)\n",
            "epoch: 18489, dloss = 4.3, gloss = -2795.4, total_loss = -277.4 (0.69)\n",
            "epoch: 18490, dloss = 6.7, gloss = -2799.1, total_loss = -276.6 (0.69)\n",
            "epoch: 18491, dloss = 5.8, gloss = -2827.3, total_loss = -279.8 (0.71)\n",
            "epoch: 18492, dloss = 4.4, gloss = -2832.7, total_loss = -281.0 (0.69)\n",
            "epoch: 18493, dloss = 6.0, gloss = -2804.5, total_loss = -277.5 (0.68)\n",
            "epoch: 18494, dloss = 6.5, gloss = -2817.3, total_loss = -278.5 (0.70)\n",
            "epoch: 18495, dloss = 4.9, gloss = -2803.3, total_loss = -277.9 (0.71)\n",
            "epoch: 18496, dloss = 4.8, gloss = -2811.4, total_loss = -278.7 (0.68)\n",
            "epoch: 18497, dloss = 5.2, gloss = -2782.8, total_loss = -275.7 (0.69)\n",
            "epoch: 18498, dloss = 7.3, gloss = -2793.4, total_loss = -275.7 (0.69)\n",
            "epoch: 18499, dloss = 6.8, gloss = -2762.2, total_loss = -272.8 (0.68)\n",
            "epoch: 18500, dloss = 5.0, gloss = -2798.2, total_loss = -277.3 (0.66)\n",
            "epoch: 18501, dloss = 5.5, gloss = -2794.3, total_loss = -276.7 (0.67)\n",
            "epoch: 18502, dloss = 3.9, gloss = -2779.5, total_loss = -276.0 (0.68)\n",
            "epoch: 18503, dloss = 4.2, gloss = -2785.6, total_loss = -276.5 (0.86)\n",
            "epoch: 18504, dloss = 5.7, gloss = -2746.9, total_loss = -271.8 (0.86)\n",
            "epoch: 18505, dloss = 4.8, gloss = -2766.1, total_loss = -274.2 (0.68)\n",
            "epoch: 18506, dloss = 6.5, gloss = -2760.3, total_loss = -272.8 (0.67)\n",
            "epoch: 18507, dloss = 5.4, gloss = -2771.4, total_loss = -274.5 (0.68)\n",
            "epoch: 18508, dloss = 3.7, gloss = -2783.3, total_loss = -276.5 (0.70)\n",
            "epoch: 18509, dloss = 5.5, gloss = -2762.4, total_loss = -273.5 (0.68)\n",
            "epoch: 18510, dloss = 3.7, gloss = -2765.8, total_loss = -274.8 (0.68)\n",
            "epoch: 18511, dloss = 3.5, gloss = -2803.7, total_loss = -278.6 (0.70)\n",
            "epoch: 18512, dloss = 6.4, gloss = -2827.1, total_loss = -279.5 (0.68)\n",
            "epoch: 18513, dloss = 4.2, gloss = -2792.1, total_loss = -277.1 (0.69)\n",
            "epoch: 18514, dloss = 4.4, gloss = -2805.3, total_loss = -278.3 (0.69)\n",
            "epoch: 18515, dloss = 2.5, gloss = -2798.7, total_loss = -278.6 (0.67)\n",
            "epoch: 18516, dloss = 3.0, gloss = -2790.9, total_loss = -277.6 (0.69)\n",
            "epoch: 18517, dloss = 4.6, gloss = -2831.5, total_loss = -280.9 (0.70)\n",
            "epoch: 18518, dloss = 4.5, gloss = -2778.2, total_loss = -275.6 (0.70)\n",
            "epoch: 18519, dloss = 5.6, gloss = -2817.5, total_loss = -279.0 (1.01)\n",
            "epoch: 18520, dloss = 4.2, gloss = -2816.0, total_loss = -279.5 (0.68)\n",
            "epoch: 18521, dloss = 2.4, gloss = -2871.6, total_loss = -285.9 (0.68)\n",
            "epoch: 18522, dloss = 4.1, gloss = -2843.2, total_loss = -282.3 (0.67)\n",
            "epoch: 18523, dloss = 7.6, gloss = -2827.5, total_loss = -278.9 (0.68)\n",
            "epoch: 18524, dloss = 8.1, gloss = -2833.6, total_loss = -279.3 (0.67)\n",
            "epoch: 18525, dloss = 2.5, gloss = -2823.7, total_loss = -281.1 (0.67)\n",
            "epoch: 18526, dloss = 5.4, gloss = -2863.5, total_loss = -283.7 (0.69)\n",
            "epoch: 18527, dloss = 1.8, gloss = -2817.2, total_loss = -280.8 (0.69)\n",
            "epoch: 18528, dloss = 1.8, gloss = -2834.7, total_loss = -282.6 (0.67)\n",
            "epoch: 18529, dloss = 5.9, gloss = -2885.6, total_loss = -285.6 (0.68)\n",
            "epoch: 18530, dloss = 4.7, gloss = -2847.1, total_loss = -282.4 (0.69)\n",
            "epoch: 18531, dloss = 3.0, gloss = -2813.8, total_loss = -279.9 (0.70)\n",
            "epoch: 18532, dloss = 5.0, gloss = -2800.1, total_loss = -277.5 (0.70)\n",
            "epoch: 18533, dloss = 5.5, gloss = -2796.9, total_loss = -277.0 (0.68)\n",
            "epoch: 18534, dloss = 4.7, gloss = -2783.3, total_loss = -276.0 (0.82)\n",
            "epoch: 18535, dloss = 3.8, gloss = -2808.4, total_loss = -278.9 (0.92)\n",
            "epoch: 18536, dloss = 4.9, gloss = -2787.6, total_loss = -276.3 (0.67)\n",
            "epoch: 18537, dloss = 4.7, gloss = -2776.2, total_loss = -275.3 (0.67)\n",
            "epoch: 18538, dloss = 2.7, gloss = -2734.0, total_loss = -272.0 (0.66)\n",
            "epoch: 18539, dloss = 4.2, gloss = -2790.9, total_loss = -277.0 (0.69)\n",
            "epoch: 18540, dloss = 2.5, gloss = -2743.6, total_loss = -273.1 (0.68)\n",
            "epoch: 18541, dloss = 6.9, gloss = -2777.0, total_loss = -274.3 (0.68)\n",
            "epoch: 18542, dloss = 5.5, gloss = -2781.3, total_loss = -275.4 (0.68)\n",
            "epoch: 18543, dloss = 5.3, gloss = -2814.9, total_loss = -278.8 (0.68)\n",
            "epoch: 18544, dloss = 5.6, gloss = -2801.4, total_loss = -277.3 (0.68)\n",
            "epoch: 18545, dloss = 3.5, gloss = -2844.8, total_loss = -282.7 (0.69)\n",
            "epoch: 18546, dloss = 3.8, gloss = -2833.2, total_loss = -281.4 (0.69)\n",
            "epoch: 18547, dloss = 3.4, gloss = -2849.7, total_loss = -283.3 (0.68)\n",
            "epoch: 18548, dloss = 5.6, gloss = -2863.7, total_loss = -283.6 (0.71)\n",
            "epoch: 18549, dloss = 4.6, gloss = -2843.3, total_loss = -282.0 (0.70)\n",
            "epoch: 18550, dloss = 4.1, gloss = -2838.5, total_loss = -281.8 (0.95)\n",
            "epoch: 18551, dloss = 3.8, gloss = -2816.0, total_loss = -279.7 (0.75)\n",
            "epoch: 18552, dloss = 3.6, gloss = -2824.3, total_loss = -280.6 (0.69)\n",
            "epoch: 18553, dloss = 5.0, gloss = -2810.3, total_loss = -278.5 (0.69)\n",
            "epoch: 18554, dloss = 5.6, gloss = -2819.9, total_loss = -279.2 (0.67)\n",
            "epoch: 18555, dloss = 4.7, gloss = -2779.8, total_loss = -275.7 (0.67)\n",
            "epoch: 18556, dloss = 4.6, gloss = -2746.9, total_loss = -272.4 (0.69)\n",
            "epoch: 18557, dloss = 4.6, gloss = -2757.2, total_loss = -273.4 (0.68)\n",
            "epoch: 18558, dloss = 12.5, gloss = -2734.1, total_loss = -267.2 (0.68)\n",
            "epoch: 18559, dloss = 3.6, gloss = -2734.8, total_loss = -271.7 (0.70)\n",
            "epoch: 18560, dloss = 2.3, gloss = -2750.0, total_loss = -273.8 (0.68)\n",
            "epoch: 18561, dloss = 9.6, gloss = -2752.9, total_loss = -270.5 (0.71)\n",
            "epoch: 18562, dloss = 3.7, gloss = -2741.4, total_loss = -272.3 (0.69)\n",
            "epoch: 18563, dloss = 4.2, gloss = -2752.3, total_loss = -273.1 (0.66)\n",
            "epoch: 18564, dloss = 4.2, gloss = -2740.8, total_loss = -272.0 (0.67)\n",
            "epoch: 18565, dloss = 3.4, gloss = -2751.9, total_loss = -273.5 (0.70)\n",
            "epoch: 18566, dloss = 3.2, gloss = -2799.2, total_loss = -278.3 (1.03)\n",
            "epoch: 18567, dloss = 6.8, gloss = -2747.8, total_loss = -271.4 (0.68)\n",
            "epoch: 18568, dloss = 4.6, gloss = -2750.9, total_loss = -272.8 (0.68)\n",
            "epoch: 18569, dloss = 3.4, gloss = -2737.0, total_loss = -272.0 (0.68)\n",
            "epoch: 18570, dloss = 4.5, gloss = -2799.0, total_loss = -277.7 (0.67)\n",
            "epoch: 18571, dloss = 5.4, gloss = -2786.5, total_loss = -275.9 (0.69)\n",
            "epoch: 18572, dloss = 4.9, gloss = -2774.4, total_loss = -275.0 (0.68)\n",
            "epoch: 18573, dloss = 5.3, gloss = -2737.9, total_loss = -271.1 (0.65)\n",
            "epoch: 18574, dloss = 4.2, gloss = -2741.4, total_loss = -272.0 (0.69)\n",
            "epoch: 18575, dloss = 3.1, gloss = -2757.5, total_loss = -274.2 (0.68)\n",
            "epoch: 18576, dloss = 2.4, gloss = -2753.1, total_loss = -274.1 (0.68)\n",
            "epoch: 18577, dloss = 3.8, gloss = -2780.1, total_loss = -276.1 (0.66)\n",
            "epoch: 18578, dloss = 6.5, gloss = -2747.3, total_loss = -271.5 (0.68)\n",
            "epoch: 18579, dloss = 3.1, gloss = -2757.1, total_loss = -274.2 (0.66)\n",
            "epoch: 18580, dloss = 7.6, gloss = -2741.3, total_loss = -270.3 (0.69)\n",
            "epoch: 18581, dloss = 4.7, gloss = -2750.4, total_loss = -272.7 (0.83)\n",
            "epoch: 18582, dloss = 4.1, gloss = -2730.8, total_loss = -271.0 (0.91)\n",
            "epoch: 18583, dloss = 4.2, gloss = -2720.2, total_loss = -269.9 (0.68)\n",
            "epoch: 18584, dloss = 12.0, gloss = -2734.4, total_loss = -267.4 (0.68)\n",
            "epoch: 18585, dloss = 3.5, gloss = -2736.9, total_loss = -272.0 (0.71)\n",
            "epoch: 18586, dloss = 4.4, gloss = -2769.8, total_loss = -274.8 (0.69)\n",
            "epoch: 18587, dloss = 4.0, gloss = -2761.2, total_loss = -274.1 (0.68)\n",
            "epoch: 18588, dloss = 5.2, gloss = -2736.7, total_loss = -271.1 (0.70)\n",
            "epoch: 18589, dloss = 3.8, gloss = -2736.4, total_loss = -271.7 (1.26)\n",
            "epoch: 18590, dloss = 4.0, gloss = -2738.5, total_loss = -271.9 (1.14)\n",
            "epoch: 18591, dloss = 5.6, gloss = -2703.1, total_loss = -267.5 (1.24)\n",
            "epoch: 18592, dloss = 8.6, gloss = -2727.0, total_loss = -268.4 (1.19)\n",
            "epoch: 18593, dloss = 5.6, gloss = -2712.7, total_loss = -268.5 (1.32)\n",
            "epoch: 18594, dloss = 5.9, gloss = -2740.6, total_loss = -271.1 (1.32)\n",
            "epoch: 18595, dloss = 5.0, gloss = -2758.3, total_loss = -273.3 (0.68)\n",
            "epoch: 18596, dloss = 3.9, gloss = -2758.7, total_loss = -273.9 (0.67)\n",
            "epoch: 18597, dloss = 6.0, gloss = -2770.8, total_loss = -274.1 (0.71)\n",
            "epoch: 18598, dloss = 2.0, gloss = -2750.0, total_loss = -274.0 (0.68)\n",
            "epoch: 18599, dloss = 2.1, gloss = -2775.5, total_loss = -276.5 (0.70)\n",
            "epoch: 18600, dloss = 4.2, gloss = -2755.0, total_loss = -273.4 (0.69)\n",
            "epoch: 18601, dloss = 5.4, gloss = -2748.1, total_loss = -272.1 (0.68)\n",
            "epoch: 18602, dloss = 2.7, gloss = -2705.3, total_loss = -269.2 (0.68)\n",
            "epoch: 18603, dloss = 5.0, gloss = -2715.6, total_loss = -269.1 (0.67)\n",
            "epoch: 18604, dloss = 4.8, gloss = -2686.2, total_loss = -266.2 (0.67)\n",
            "epoch: 18605, dloss = 4.7, gloss = -2701.5, total_loss = -267.8 (0.68)\n",
            "epoch: 18606, dloss = 4.4, gloss = -2713.2, total_loss = -269.1 (0.69)\n",
            "epoch: 18607, dloss = 4.3, gloss = -2701.4, total_loss = -268.0 (0.70)\n",
            "epoch: 18608, dloss = 2.8, gloss = -2739.7, total_loss = -272.6 (0.68)\n",
            "epoch: 18609, dloss = 3.3, gloss = -2745.5, total_loss = -272.9 (0.83)\n",
            "epoch: 18610, dloss = 5.6, gloss = -2699.4, total_loss = -267.1 (0.91)\n",
            "epoch: 18611, dloss = 2.4, gloss = -2746.4, total_loss = -273.4 (0.69)\n",
            "epoch: 18612, dloss = 5.3, gloss = -2744.3, total_loss = -271.8 (0.68)\n",
            "epoch: 18613, dloss = 6.0, gloss = -2728.6, total_loss = -269.9 (0.69)\n",
            "epoch: 18614, dloss = 2.7, gloss = -2736.1, total_loss = -272.2 (0.70)\n",
            "epoch: 18615, dloss = 5.0, gloss = -2707.5, total_loss = -268.2 (0.70)\n",
            "epoch: 18616, dloss = 3.5, gloss = -2712.2, total_loss = -269.5 (0.71)\n",
            "epoch: 18617, dloss = 3.0, gloss = -2728.7, total_loss = -271.4 (0.72)\n",
            "epoch: 18618, dloss = 4.5, gloss = -2763.4, total_loss = -274.1 (0.67)\n",
            "epoch: 18619, dloss = 3.6, gloss = -2760.9, total_loss = -274.3 (0.68)\n",
            "epoch: 18620, dloss = 4.7, gloss = -2786.0, total_loss = -276.2 (0.68)\n",
            "epoch: 18621, dloss = 4.5, gloss = -2771.4, total_loss = -274.9 (0.70)\n",
            "epoch: 18622, dloss = 4.7, gloss = -2814.6, total_loss = -279.1 (0.68)\n",
            "epoch: 18623, dloss = 3.1, gloss = -2794.8, total_loss = -277.9 (0.68)\n",
            "epoch: 18624, dloss = 3.6, gloss = -2770.1, total_loss = -275.2 (0.68)\n",
            "epoch: 18625, dloss = 5.1, gloss = -2794.0, total_loss = -276.8 (1.01)\n",
            "epoch: 18626, dloss = 3.9, gloss = -2779.3, total_loss = -276.0 (0.70)\n",
            "epoch: 18627, dloss = 4.9, gloss = -2757.5, total_loss = -273.3 (0.68)\n",
            "epoch: 18628, dloss = 2.3, gloss = -2756.9, total_loss = -274.5 (0.70)\n",
            "epoch: 18629, dloss = 4.1, gloss = -2734.3, total_loss = -271.4 (0.71)\n",
            "epoch: 18630, dloss = 6.1, gloss = -2766.0, total_loss = -273.5 (0.68)\n",
            "epoch: 18631, dloss = 5.9, gloss = -2753.4, total_loss = -272.4 (0.71)\n",
            "epoch: 18632, dloss = 5.2, gloss = -2746.7, total_loss = -272.1 (0.67)\n",
            "epoch: 18633, dloss = 4.3, gloss = -2778.2, total_loss = -275.7 (0.69)\n",
            "epoch: 18634, dloss = 3.9, gloss = -2776.3, total_loss = -275.7 (0.69)\n",
            "epoch: 18635, dloss = 7.2, gloss = -2790.8, total_loss = -275.5 (0.69)\n",
            "epoch: 18636, dloss = 4.9, gloss = -2761.0, total_loss = -273.6 (0.66)\n",
            "epoch: 18637, dloss = 4.1, gloss = -2765.0, total_loss = -274.4 (0.67)\n",
            "epoch: 18638, dloss = 4.1, gloss = -2752.1, total_loss = -273.2 (0.66)\n",
            "epoch: 18639, dloss = 4.3, gloss = -2743.1, total_loss = -272.2 (0.68)\n",
            "epoch: 18640, dloss = 4.9, gloss = -2754.3, total_loss = -273.0 (0.82)\n",
            "epoch: 18641, dloss = 5.2, gloss = -2775.7, total_loss = -275.0 (0.90)\n",
            "epoch: 18642, dloss = 2.7, gloss = -2784.0, total_loss = -277.0 (0.69)\n",
            "epoch: 18643, dloss = 7.1, gloss = -2769.1, total_loss = -273.4 (0.71)\n",
            "epoch: 18644, dloss = 4.3, gloss = -2778.7, total_loss = -275.7 (0.70)\n",
            "epoch: 18645, dloss = 5.0, gloss = -2828.8, total_loss = -280.4 (0.68)\n",
            "epoch: 18646, dloss = 4.8, gloss = -2793.8, total_loss = -277.0 (0.69)\n",
            "epoch: 18647, dloss = 8.0, gloss = -2809.4, total_loss = -277.0 (0.68)\n",
            "epoch: 18648, dloss = 4.4, gloss = -2807.6, total_loss = -278.6 (0.68)\n",
            "epoch: 18649, dloss = 3.6, gloss = -2781.0, total_loss = -276.3 (0.68)\n",
            "epoch: 18650, dloss = 3.7, gloss = -2798.6, total_loss = -278.0 (0.70)\n",
            "epoch: 18651, dloss = 5.4, gloss = -2795.6, total_loss = -276.9 (0.68)\n",
            "epoch: 18652, dloss = 4.2, gloss = -2809.3, total_loss = -278.8 (0.68)\n",
            "epoch: 18653, dloss = 3.2, gloss = -2789.7, total_loss = -277.4 (0.69)\n",
            "epoch: 18654, dloss = 3.5, gloss = -2766.5, total_loss = -274.9 (0.66)\n",
            "epoch: 18655, dloss = 3.7, gloss = -2776.5, total_loss = -275.8 (0.66)\n",
            "epoch: 18656, dloss = 4.4, gloss = -2771.7, total_loss = -275.0 (0.95)\n",
            "epoch: 18657, dloss = 5.3, gloss = -2783.6, total_loss = -275.7 (0.79)\n",
            "epoch: 18658, dloss = 4.6, gloss = -2752.1, total_loss = -272.9 (0.68)\n",
            "epoch: 18659, dloss = 2.3, gloss = -2758.6, total_loss = -274.7 (0.67)\n",
            "epoch: 18660, dloss = 4.7, gloss = -2775.1, total_loss = -275.2 (0.67)\n",
            "epoch: 18661, dloss = 4.3, gloss = -2830.9, total_loss = -280.9 (0.65)\n",
            "epoch: 18662, dloss = 3.7, gloss = -2780.2, total_loss = -276.2 (0.66)\n",
            "epoch: 18663, dloss = 2.4, gloss = -2793.5, total_loss = -278.2 (0.70)\n",
            "epoch: 18664, dloss = 4.5, gloss = -2780.4, total_loss = -275.8 (0.68)\n",
            "epoch: 18665, dloss = 5.0, gloss = -2805.2, total_loss = -278.0 (0.66)\n",
            "epoch: 18666, dloss = 4.6, gloss = -2792.3, total_loss = -276.9 (0.68)\n",
            "epoch: 18667, dloss = 3.6, gloss = -2805.7, total_loss = -278.8 (0.69)\n",
            "epoch: 18668, dloss = 12.8, gloss = -2782.4, total_loss = -271.9 (0.69)\n",
            "epoch: 18669, dloss = 5.6, gloss = -2777.3, total_loss = -274.9 (0.68)\n",
            "epoch: 18670, dloss = 6.1, gloss = -2788.3, total_loss = -275.8 (0.70)\n",
            "epoch: 18671, dloss = 7.1, gloss = -2867.3, total_loss = -283.2 (0.72)\n",
            "epoch: 18672, dloss = 6.2, gloss = -2832.2, total_loss = -280.1 (1.01)\n",
            "epoch: 18673, dloss = 4.6, gloss = -2814.0, total_loss = -279.1 (0.71)\n",
            "epoch: 18674, dloss = 3.8, gloss = -2798.5, total_loss = -278.0 (0.70)\n",
            "epoch: 18675, dloss = 4.8, gloss = -2799.0, total_loss = -277.5 (0.69)\n",
            "epoch: 18676, dloss = 4.3, gloss = -2833.3, total_loss = -281.2 (0.70)\n",
            "epoch: 18677, dloss = 3.8, gloss = -2832.9, total_loss = -281.4 (0.69)\n",
            "epoch: 18678, dloss = 4.7, gloss = -2867.0, total_loss = -284.4 (0.68)\n",
            "epoch: 18679, dloss = -0.4, gloss = -2843.0, total_loss = -284.5 (0.69)\n",
            "epoch: 18680, dloss = 5.6, gloss = -2874.3, total_loss = -284.6 (0.69)\n",
            "epoch: 18681, dloss = 6.2, gloss = -2890.0, total_loss = -285.9 (0.69)\n",
            "epoch: 18682, dloss = 5.5, gloss = -2840.5, total_loss = -281.3 (0.68)\n",
            "epoch: 18683, dloss = 4.2, gloss = -2854.3, total_loss = -283.3 (0.69)\n",
            "epoch: 18684, dloss = 17.1, gloss = -2854.7, total_loss = -276.9 (0.69)\n",
            "epoch: 18685, dloss = 4.4, gloss = -2849.3, total_loss = -282.7 (0.69)\n",
            "epoch: 18686, dloss = 2.9, gloss = -2828.2, total_loss = -281.4 (0.69)\n",
            "epoch: 18687, dloss = 4.5, gloss = -2828.2, total_loss = -280.6 (0.90)\n",
            "epoch: 18688, dloss = 1.1, gloss = -2869.5, total_loss = -286.4 (0.84)\n",
            "epoch: 18689, dloss = 6.7, gloss = -2894.2, total_loss = -286.1 (0.71)\n",
            "epoch: 18690, dloss = 5.3, gloss = -2886.0, total_loss = -286.0 (0.69)\n",
            "epoch: 18691, dloss = 4.7, gloss = -2889.9, total_loss = -286.7 (0.69)\n",
            "epoch: 18692, dloss = 4.5, gloss = -2867.9, total_loss = -284.5 (0.68)\n",
            "epoch: 18693, dloss = 2.3, gloss = -2873.1, total_loss = -286.2 (0.68)\n",
            "epoch: 18694, dloss = 7.0, gloss = -2861.1, total_loss = -282.6 (0.69)\n",
            "epoch: 18695, dloss = 3.8, gloss = -2850.8, total_loss = -283.2 (0.69)\n",
            "epoch: 18696, dloss = 3.9, gloss = -2843.3, total_loss = -282.4 (0.67)\n",
            "epoch: 18697, dloss = 4.1, gloss = -2859.6, total_loss = -283.9 (0.67)\n",
            "epoch: 18698, dloss = 4.5, gloss = -2844.9, total_loss = -282.3 (0.68)\n",
            "epoch: 18699, dloss = 10.5, gloss = -2852.4, total_loss = -280.0 (0.70)\n",
            "epoch: 18700, dloss = 4.4, gloss = -2832.9, total_loss = -281.1 (0.68)\n",
            "epoch: 18701, dloss = 2.8, gloss = -2839.5, total_loss = -282.6 (0.68)\n",
            "epoch: 18702, dloss = 5.4, gloss = -2845.3, total_loss = -281.8 (0.73)\n",
            "epoch: 18703, dloss = 5.6, gloss = -2816.3, total_loss = -278.9 (1.00)\n",
            "epoch: 18704, dloss = 4.3, gloss = -2853.3, total_loss = -283.2 (0.67)\n",
            "epoch: 18705, dloss = 5.0, gloss = -2848.5, total_loss = -282.4 (0.69)\n",
            "epoch: 18706, dloss = 5.7, gloss = -2835.2, total_loss = -280.7 (0.69)\n",
            "epoch: 18707, dloss = 4.8, gloss = -2872.8, total_loss = -284.9 (0.70)\n",
            "epoch: 18708, dloss = 6.1, gloss = -2845.8, total_loss = -281.5 (0.68)\n",
            "epoch: 18709, dloss = 3.2, gloss = -2872.0, total_loss = -285.6 (0.67)\n",
            "epoch: 18710, dloss = 6.6, gloss = -2856.2, total_loss = -282.3 (0.68)\n",
            "epoch: 18711, dloss = 4.6, gloss = -2859.2, total_loss = -283.6 (0.66)\n",
            "epoch: 18712, dloss = 2.3, gloss = -2871.4, total_loss = -286.0 (0.67)\n",
            "epoch: 18713, dloss = 3.6, gloss = -2885.8, total_loss = -286.8 (0.68)\n",
            "epoch: 18714, dloss = 4.3, gloss = -2832.2, total_loss = -281.1 (0.68)\n",
            "epoch: 18715, dloss = 3.3, gloss = -2844.9, total_loss = -282.9 (0.68)\n",
            "epoch: 18716, dloss = 5.2, gloss = -2868.0, total_loss = -284.2 (0.72)\n",
            "epoch: 18717, dloss = 4.6, gloss = -2852.6, total_loss = -282.9 (0.68)\n",
            "epoch: 18718, dloss = 4.4, gloss = -2837.1, total_loss = -281.5 (0.84)\n",
            "epoch: 18719, dloss = 4.1, gloss = -2878.1, total_loss = -285.7 (0.86)\n",
            "epoch: 18720, dloss = 4.7, gloss = -2873.3, total_loss = -285.0 (0.69)\n",
            "epoch: 18721, dloss = 4.6, gloss = -2856.6, total_loss = -283.3 (0.68)\n",
            "epoch: 18722, dloss = 6.3, gloss = -2834.1, total_loss = -280.3 (0.69)\n",
            "epoch: 18723, dloss = 3.4, gloss = -2818.6, total_loss = -280.2 (0.69)\n",
            "epoch: 18724, dloss = 4.7, gloss = -2847.0, total_loss = -282.3 (0.69)\n",
            "epoch: 18725, dloss = 4.4, gloss = -2830.6, total_loss = -280.9 (0.71)\n",
            "epoch: 18726, dloss = 3.2, gloss = -2862.0, total_loss = -284.6 (0.70)\n",
            "epoch: 18727, dloss = 4.2, gloss = -2870.8, total_loss = -285.0 (0.68)\n",
            "epoch: 18728, dloss = 5.0, gloss = -2859.3, total_loss = -283.4 (0.82)\n",
            "epoch: 18729, dloss = 4.9, gloss = -2869.5, total_loss = -284.5 (1.00)\n",
            "epoch: 18730, dloss = 5.8, gloss = -2861.1, total_loss = -283.2 (0.70)\n",
            "epoch: 18731, dloss = 7.5, gloss = -2894.5, total_loss = -285.7 (0.67)\n",
            "epoch: 18732, dloss = 4.5, gloss = -2881.9, total_loss = -285.9 (0.70)\n",
            "epoch: 18733, dloss = 6.0, gloss = -2901.1, total_loss = -287.1 (0.88)\n",
            "epoch: 18734, dloss = 5.2, gloss = -2902.9, total_loss = -287.7 (0.85)\n",
            "epoch: 18735, dloss = 4.1, gloss = -2872.1, total_loss = -285.2 (0.67)\n",
            "epoch: 18736, dloss = 4.7, gloss = -2874.3, total_loss = -285.1 (0.66)\n",
            "epoch: 18737, dloss = 4.0, gloss = -2860.2, total_loss = -284.0 (0.69)\n",
            "epoch: 18738, dloss = 4.5, gloss = -2872.0, total_loss = -285.0 (0.68)\n",
            "epoch: 18739, dloss = 4.7, gloss = -2860.1, total_loss = -283.7 (0.68)\n",
            "epoch: 18740, dloss = 4.8, gloss = -2855.7, total_loss = -283.2 (0.70)\n",
            "epoch: 18741, dloss = 3.4, gloss = -2870.2, total_loss = -285.3 (0.69)\n",
            "epoch: 18742, dloss = 1.9, gloss = -2876.9, total_loss = -286.8 (0.68)\n",
            "epoch: 18743, dloss = 4.8, gloss = -2880.9, total_loss = -285.7 (0.68)\n",
            "epoch: 18744, dloss = 4.4, gloss = -2909.9, total_loss = -288.8 (0.69)\n",
            "epoch: 18745, dloss = 5.3, gloss = -2874.1, total_loss = -284.8 (0.67)\n",
            "epoch: 18746, dloss = 5.1, gloss = -2918.4, total_loss = -289.3 (0.69)\n",
            "epoch: 18747, dloss = 3.6, gloss = -2926.2, total_loss = -290.8 (0.69)\n",
            "epoch: 18748, dloss = 9.8, gloss = -2893.0, total_loss = -284.4 (0.68)\n",
            "epoch: 18749, dloss = 2.7, gloss = -2896.6, total_loss = -288.3 (0.98)\n",
            "epoch: 18750, dloss = 6.6, gloss = -2911.0, total_loss = -287.8 (0.74)\n",
            "epoch: 18751, dloss = 5.8, gloss = -2903.4, total_loss = -287.4 (0.66)\n",
            "epoch: 18752, dloss = 3.4, gloss = -2900.4, total_loss = -288.3 (0.69)\n",
            "epoch: 18753, dloss = 5.4, gloss = -2920.8, total_loss = -289.4 (0.67)\n",
            "epoch: 18754, dloss = 3.0, gloss = -2924.5, total_loss = -290.9 (0.69)\n",
            "epoch: 18755, dloss = 5.3, gloss = -2924.3, total_loss = -289.8 (0.69)\n",
            "epoch: 18756, dloss = 4.9, gloss = -2929.3, total_loss = -290.5 (0.69)\n",
            "epoch: 18757, dloss = 3.8, gloss = -2908.2, total_loss = -288.9 (0.71)\n",
            "epoch: 18758, dloss = 5.0, gloss = -2920.8, total_loss = -289.6 (0.69)\n",
            "epoch: 18759, dloss = 4.7, gloss = -2942.0, total_loss = -291.9 (0.69)\n",
            "epoch: 18760, dloss = 5.9, gloss = -2942.7, total_loss = -291.3 (0.69)\n",
            "epoch: 18761, dloss = 5.2, gloss = -2931.0, total_loss = -290.5 (0.70)\n",
            "epoch: 18762, dloss = 4.6, gloss = -2932.4, total_loss = -290.9 (0.72)\n",
            "epoch: 18763, dloss = 4.2, gloss = -2917.0, total_loss = -289.6 (0.67)\n",
            "epoch: 18764, dloss = 5.6, gloss = -2928.3, total_loss = -290.0 (0.82)\n",
            "epoch: 18765, dloss = 4.5, gloss = -2930.4, total_loss = -290.8 (0.91)\n",
            "epoch: 18766, dloss = 3.9, gloss = -2905.8, total_loss = -288.6 (0.69)\n",
            "epoch: 18767, dloss = 4.9, gloss = -2920.5, total_loss = -289.6 (0.67)\n",
            "epoch: 18768, dloss = 8.4, gloss = -2941.3, total_loss = -289.9 (0.69)\n",
            "epoch: 18769, dloss = 4.7, gloss = -2928.9, total_loss = -290.6 (0.69)\n",
            "epoch: 18770, dloss = 5.2, gloss = -2955.3, total_loss = -292.9 (0.69)\n",
            "epoch: 18771, dloss = 4.5, gloss = -2936.5, total_loss = -291.4 (0.67)\n",
            "epoch: 18772, dloss = 4.8, gloss = -2938.9, total_loss = -291.5 (0.69)\n",
            "epoch: 18773, dloss = 4.6, gloss = -2934.1, total_loss = -291.1 (0.68)\n",
            "epoch: 18774, dloss = 4.5, gloss = -2942.2, total_loss = -292.0 (0.69)\n",
            "epoch: 18775, dloss = 5.0, gloss = -2935.1, total_loss = -291.0 (0.70)\n",
            "epoch: 18776, dloss = 4.3, gloss = -2936.1, total_loss = -291.5 (0.70)\n",
            "epoch: 18777, dloss = 6.3, gloss = -2962.8, total_loss = -293.1 (0.69)\n",
            "epoch: 18778, dloss = 3.3, gloss = -2960.9, total_loss = -294.4 (0.70)\n",
            "epoch: 18779, dloss = 4.7, gloss = -2956.8, total_loss = -293.3 (0.70)\n",
            "epoch: 18780, dloss = 2.5, gloss = -2938.3, total_loss = -292.6 (1.26)\n",
            "epoch: 18781, dloss = 3.9, gloss = -2968.1, total_loss = -294.9 (0.70)\n",
            "epoch: 18782, dloss = 16.5, gloss = -2957.7, total_loss = -287.5 (0.68)\n",
            "epoch: 18783, dloss = 5.4, gloss = -2968.7, total_loss = -294.2 (0.69)\n",
            "epoch: 18784, dloss = 6.8, gloss = -2965.1, total_loss = -293.1 (0.69)\n",
            "epoch: 18785, dloss = 10.9, gloss = -2967.7, total_loss = -291.3 (0.68)\n",
            "epoch: 18786, dloss = 7.2, gloss = -2977.2, total_loss = -294.1 (0.69)\n",
            "epoch: 18787, dloss = 5.3, gloss = -2966.3, total_loss = -294.0 (0.68)\n",
            "epoch: 18788, dloss = 4.4, gloss = -2958.3, total_loss = -293.6 (0.69)\n",
            "epoch: 18789, dloss = 5.8, gloss = -2943.0, total_loss = -291.4 (0.69)\n",
            "epoch: 18790, dloss = 6.0, gloss = -2924.1, total_loss = -289.4 (0.68)\n",
            "epoch: 18791, dloss = 5.0, gloss = -2941.4, total_loss = -291.6 (0.67)\n",
            "epoch: 18792, dloss = 6.0, gloss = -2937.9, total_loss = -290.8 (0.70)\n",
            "epoch: 18793, dloss = 4.9, gloss = -2944.7, total_loss = -292.0 (0.70)\n",
            "epoch: 18794, dloss = 5.3, gloss = -2939.1, total_loss = -291.3 (0.68)\n",
            "epoch: 18795, dloss = 4.8, gloss = -2955.8, total_loss = -293.2 (0.89)\n",
            "epoch: 18796, dloss = 22.5, gloss = -2948.2, total_loss = -283.6 (0.84)\n",
            "epoch: 18797, dloss = 3.2, gloss = -2955.6, total_loss = -293.9 (0.68)\n",
            "epoch: 18798, dloss = 3.9, gloss = -2949.9, total_loss = -293.1 (0.69)\n",
            "epoch: 18799, dloss = 5.8, gloss = -2974.5, total_loss = -294.6 (0.69)\n",
            "epoch: 18800, dloss = 9.6, gloss = -2971.5, total_loss = -292.3 (0.69)\n",
            "epoch: 18801, dloss = 5.0, gloss = -2948.8, total_loss = -292.4 (0.69)\n",
            "epoch: 18802, dloss = 19.9, gloss = -2990.7, total_loss = -289.1 (0.66)\n",
            "epoch: 18803, dloss = 5.7, gloss = -2997.9, total_loss = -296.9 (0.70)\n",
            "epoch: 18804, dloss = 4.0, gloss = -2955.2, total_loss = -293.5 (0.70)\n",
            "epoch: 18805, dloss = 6.4, gloss = -2965.4, total_loss = -293.3 (0.68)\n",
            "epoch: 18806, dloss = 5.3, gloss = -2950.8, total_loss = -292.4 (0.68)\n",
            "epoch: 18807, dloss = 5.5, gloss = -2954.5, total_loss = -292.7 (0.69)\n",
            "epoch: 18808, dloss = 4.1, gloss = -2950.1, total_loss = -292.9 (0.66)\n",
            "epoch: 18809, dloss = 4.3, gloss = -2957.6, total_loss = -293.6 (0.68)\n",
            "epoch: 18810, dloss = 1.9, gloss = -2965.1, total_loss = -295.6 (0.70)\n",
            "epoch: 18811, dloss = 4.9, gloss = -2973.0, total_loss = -294.8 (1.02)\n",
            "epoch: 18812, dloss = 3.9, gloss = -2946.6, total_loss = -292.7 (0.69)\n",
            "epoch: 18813, dloss = 5.3, gloss = -2969.5, total_loss = -294.3 (0.69)\n",
            "epoch: 18814, dloss = 3.8, gloss = -2936.6, total_loss = -291.8 (0.69)\n",
            "epoch: 18815, dloss = 5.1, gloss = -2951.0, total_loss = -292.6 (0.67)\n",
            "epoch: 18816, dloss = 4.6, gloss = -2956.4, total_loss = -293.3 (0.69)\n",
            "epoch: 18817, dloss = 3.6, gloss = -2963.4, total_loss = -294.6 (0.68)\n",
            "epoch: 18818, dloss = 4.8, gloss = -2960.0, total_loss = -293.6 (0.68)\n",
            "epoch: 18819, dloss = 5.6, gloss = -2963.8, total_loss = -293.6 (0.70)\n",
            "epoch: 18820, dloss = 4.4, gloss = -2971.5, total_loss = -295.0 (0.71)\n",
            "epoch: 18821, dloss = 6.7, gloss = -2979.2, total_loss = -294.6 (0.68)\n",
            "epoch: 18822, dloss = 4.2, gloss = -2977.3, total_loss = -295.6 (0.67)\n",
            "epoch: 18823, dloss = 6.4, gloss = -2955.5, total_loss = -292.4 (0.70)\n",
            "epoch: 18824, dloss = 5.1, gloss = -2968.3, total_loss = -294.3 (0.67)\n",
            "epoch: 18825, dloss = 4.9, gloss = -2966.3, total_loss = -294.2 (0.70)\n",
            "epoch: 18826, dloss = 6.1, gloss = -2998.7, total_loss = -296.8 (0.91)\n",
            "epoch: 18827, dloss = 3.9, gloss = -2991.7, total_loss = -297.2 (0.88)\n",
            "epoch: 18828, dloss = 5.4, gloss = -3012.0, total_loss = -298.5 (0.71)\n",
            "epoch: 18829, dloss = 4.4, gloss = -2992.0, total_loss = -297.0 (0.68)\n",
            "epoch: 18830, dloss = 4.9, gloss = -2989.8, total_loss = -296.5 (0.70)\n",
            "epoch: 18831, dloss = 5.6, gloss = -3003.3, total_loss = -297.6 (0.70)\n",
            "epoch: 18832, dloss = 5.4, gloss = -3010.6, total_loss = -298.4 (0.70)\n",
            "epoch: 18833, dloss = 4.5, gloss = -2985.9, total_loss = -296.4 (0.70)\n",
            "epoch: 18834, dloss = 5.0, gloss = -2975.3, total_loss = -295.0 (0.70)\n",
            "epoch: 18835, dloss = 5.4, gloss = -3006.0, total_loss = -297.9 (0.69)\n",
            "epoch: 18836, dloss = 2.7, gloss = -3028.0, total_loss = -301.5 (0.70)\n",
            "epoch: 18837, dloss = 7.9, gloss = -3025.7, total_loss = -298.6 (0.70)\n",
            "epoch: 18838, dloss = 4.7, gloss = -3015.2, total_loss = -299.1 (0.81)\n",
            "epoch: 18839, dloss = 4.6, gloss = -2987.6, total_loss = -296.5 (0.97)\n",
            "epoch: 18840, dloss = 5.8, gloss = -3028.3, total_loss = -299.9 (0.70)\n",
            "epoch: 18841, dloss = 4.3, gloss = -3021.7, total_loss = -300.0 (1.01)\n",
            "epoch: 18842, dloss = 4.1, gloss = -3019.4, total_loss = -299.9 (0.76)\n",
            "epoch: 18843, dloss = 4.9, gloss = -3006.6, total_loss = -298.2 (0.69)\n",
            "epoch: 18844, dloss = 6.8, gloss = -2988.2, total_loss = -295.4 (0.68)\n",
            "epoch: 18845, dloss = 6.4, gloss = -3017.0, total_loss = -298.5 (0.89)\n",
            "epoch: 18846, dloss = 2.1, gloss = -3038.4, total_loss = -302.8 (0.91)\n",
            "epoch: 18847, dloss = 3.1, gloss = -3015.5, total_loss = -300.0 (0.71)\n",
            "epoch: 18848, dloss = 5.9, gloss = -2983.8, total_loss = -295.4 (0.70)\n",
            "epoch: 18849, dloss = 12.7, gloss = -2993.0, total_loss = -292.9 (0.70)\n",
            "epoch: 18850, dloss = 5.0, gloss = -2987.8, total_loss = -296.3 (0.71)\n",
            "epoch: 18851, dloss = 5.9, gloss = -3005.4, total_loss = -297.6 (0.69)\n",
            "epoch: 18852, dloss = 10.4, gloss = -3005.5, total_loss = -295.3 (0.69)\n",
            "epoch: 18853, dloss = 4.5, gloss = -2990.6, total_loss = -296.8 (0.69)\n",
            "epoch: 18854, dloss = 4.7, gloss = -3011.3, total_loss = -298.8 (0.69)\n",
            "epoch: 18855, dloss = 7.3, gloss = -2986.1, total_loss = -294.9 (0.74)\n",
            "epoch: 18856, dloss = 4.3, gloss = -3008.4, total_loss = -298.7 (1.00)\n",
            "epoch: 18857, dloss = 3.7, gloss = -2995.5, total_loss = -297.7 (0.69)\n",
            "epoch: 18858, dloss = 5.2, gloss = -3008.6, total_loss = -298.3 (0.69)\n",
            "epoch: 18859, dloss = 4.1, gloss = -3009.1, total_loss = -298.9 (0.72)\n",
            "epoch: 18860, dloss = 4.7, gloss = -3003.8, total_loss = -298.0 (0.71)\n",
            "epoch: 18861, dloss = 5.6, gloss = -3002.9, total_loss = -297.5 (0.69)\n",
            "epoch: 18862, dloss = 3.8, gloss = -3007.9, total_loss = -298.9 (0.68)\n",
            "epoch: 18863, dloss = 3.7, gloss = -3006.5, total_loss = -298.8 (0.68)\n",
            "epoch: 18864, dloss = 5.0, gloss = -3011.7, total_loss = -298.7 (0.68)\n",
            "epoch: 18865, dloss = 4.4, gloss = -3000.5, total_loss = -297.9 (0.70)\n",
            "epoch: 18866, dloss = 3.9, gloss = -2984.1, total_loss = -296.5 (0.69)\n",
            "epoch: 18867, dloss = 6.5, gloss = -2992.6, total_loss = -296.0 (0.69)\n",
            "epoch: 18868, dloss = 6.1, gloss = -3030.7, total_loss = -300.0 (0.69)\n",
            "epoch: 18869, dloss = 4.8, gloss = -3024.0, total_loss = -300.0 (0.69)\n",
            "epoch: 18870, dloss = 4.7, gloss = -3022.2, total_loss = -299.9 (0.68)\n",
            "epoch: 18871, dloss = 5.3, gloss = -3037.3, total_loss = -301.1 (0.86)\n",
            "epoch: 18872, dloss = 4.5, gloss = -3027.5, total_loss = -300.5 (0.85)\n",
            "epoch: 18873, dloss = 4.8, gloss = -3019.8, total_loss = -299.6 (0.70)\n",
            "epoch: 18874, dloss = 3.5, gloss = -3000.8, total_loss = -298.3 (0.70)\n",
            "epoch: 18875, dloss = 4.5, gloss = -3011.0, total_loss = -298.8 (0.71)\n",
            "epoch: 18876, dloss = 2.8, gloss = -3032.6, total_loss = -301.9 (0.68)\n",
            "epoch: 18877, dloss = 6.6, gloss = -3027.2, total_loss = -299.4 (0.70)\n",
            "epoch: 18878, dloss = 4.9, gloss = -3008.7, total_loss = -298.4 (0.69)\n",
            "epoch: 18879, dloss = 5.7, gloss = -3003.2, total_loss = -297.5 (0.71)\n",
            "epoch: 18880, dloss = 5.3, gloss = -2993.0, total_loss = -296.7 (0.68)\n",
            "epoch: 18881, dloss = 4.6, gloss = -2982.3, total_loss = -295.9 (0.66)\n",
            "epoch: 18882, dloss = 6.0, gloss = -2988.1, total_loss = -295.8 (0.71)\n",
            "epoch: 18883, dloss = 3.6, gloss = -2979.4, total_loss = -296.1 (0.69)\n",
            "epoch: 18884, dloss = 4.2, gloss = -3020.2, total_loss = -299.9 (0.69)\n",
            "epoch: 18885, dloss = 6.0, gloss = -3019.5, total_loss = -299.0 (0.69)\n",
            "epoch: 18886, dloss = 4.1, gloss = -2991.5, total_loss = -297.1 (0.77)\n",
            "epoch: 18887, dloss = 5.7, gloss = -2989.5, total_loss = -296.1 (0.98)\n",
            "epoch: 18888, dloss = 4.9, gloss = -2983.1, total_loss = -295.9 (0.69)\n",
            "epoch: 18889, dloss = 6.2, gloss = -2986.0, total_loss = -295.5 (0.69)\n",
            "epoch: 18890, dloss = 3.7, gloss = -2977.2, total_loss = -295.9 (0.68)\n",
            "epoch: 18891, dloss = 6.3, gloss = -2985.9, total_loss = -295.5 (0.66)\n",
            "epoch: 18892, dloss = 6.6, gloss = -3016.2, total_loss = -298.3 (0.68)\n",
            "epoch: 18893, dloss = 8.7, gloss = -3076.0, total_loss = -303.3 (0.69)\n",
            "epoch: 18894, dloss = 8.4, gloss = -3081.2, total_loss = -303.9 (0.66)\n",
            "epoch: 18895, dloss = 7.4, gloss = -3073.5, total_loss = -303.7 (0.69)\n",
            "epoch: 18896, dloss = 7.5, gloss = -3092.5, total_loss = -305.5 (0.68)\n",
            "epoch: 18897, dloss = 6.8, gloss = -3099.6, total_loss = -306.6 (0.68)\n",
            "epoch: 18898, dloss = 8.7, gloss = -3138.1, total_loss = -309.5 (0.67)\n",
            "epoch: 18899, dloss = 7.0, gloss = -3119.9, total_loss = -308.5 (0.68)\n",
            "epoch: 18900, dloss = 6.3, gloss = -3132.9, total_loss = -310.1 (0.69)\n",
            "epoch: 18901, dloss = 7.2, gloss = -3165.9, total_loss = -313.0 (0.68)\n",
            "epoch: 18902, dloss = 6.1, gloss = -3182.7, total_loss = -315.2 (0.87)\n",
            "epoch: 18903, dloss = 7.1, gloss = -3177.8, total_loss = -314.2 (0.87)\n",
            "epoch: 18904, dloss = 6.9, gloss = -3174.1, total_loss = -314.0 (0.71)\n",
            "epoch: 18905, dloss = 6.3, gloss = -3166.8, total_loss = -313.6 (0.67)\n",
            "epoch: 18906, dloss = 7.5, gloss = -3179.7, total_loss = -314.2 (0.69)\n",
            "epoch: 18907, dloss = 6.5, gloss = -3165.2, total_loss = -313.3 (0.69)\n",
            "epoch: 18908, dloss = 5.9, gloss = -3173.2, total_loss = -314.4 (0.70)\n",
            "epoch: 18909, dloss = 9.4, gloss = -3177.8, total_loss = -313.1 (0.68)\n",
            "epoch: 18910, dloss = 7.0, gloss = -3183.3, total_loss = -314.9 (0.69)\n",
            "epoch: 18911, dloss = 7.0, gloss = -3144.5, total_loss = -310.9 (0.67)\n",
            "epoch: 18912, dloss = 7.5, gloss = -3125.6, total_loss = -308.8 (0.69)\n",
            "epoch: 18913, dloss = 5.9, gloss = -3150.2, total_loss = -312.1 (0.69)\n",
            "epoch: 18914, dloss = 5.7, gloss = -3135.2, total_loss = -310.7 (0.70)\n",
            "epoch: 18915, dloss = 5.6, gloss = -3155.0, total_loss = -312.7 (0.72)\n",
            "epoch: 18916, dloss = 6.8, gloss = -3145.7, total_loss = -311.2 (0.70)\n",
            "epoch: 18917, dloss = 6.2, gloss = -3113.5, total_loss = -308.3 (0.68)\n",
            "epoch: 18918, dloss = 5.5, gloss = -3074.2, total_loss = -304.7 (1.05)\n",
            "epoch: 18919, dloss = 5.5, gloss = -3075.9, total_loss = -304.9 (0.67)\n",
            "epoch: 18920, dloss = 6.0, gloss = -3102.2, total_loss = -307.2 (0.68)\n",
            "epoch: 18921, dloss = 5.7, gloss = -3134.0, total_loss = -310.6 (0.69)\n",
            "epoch: 18922, dloss = 7.0, gloss = -3147.1, total_loss = -311.2 (0.70)\n",
            "epoch: 18923, dloss = 5.9, gloss = -3140.2, total_loss = -311.1 (0.69)\n",
            "epoch: 18924, dloss = 6.6, gloss = -3153.7, total_loss = -312.1 (0.67)\n",
            "epoch: 18925, dloss = 7.2, gloss = -3143.2, total_loss = -310.7 (0.68)\n",
            "epoch: 18926, dloss = 5.7, gloss = -3158.6, total_loss = -313.0 (0.68)\n",
            "epoch: 18927, dloss = 4.7, gloss = -3174.6, total_loss = -315.1 (0.67)\n",
            "epoch: 18928, dloss = 6.9, gloss = -3171.0, total_loss = -313.6 (0.68)\n",
            "epoch: 18929, dloss = 6.7, gloss = -3162.1, total_loss = -312.9 (0.69)\n",
            "epoch: 18930, dloss = 7.8, gloss = -3173.9, total_loss = -313.5 (0.71)\n",
            "epoch: 18931, dloss = 8.8, gloss = -3157.1, total_loss = -311.3 (0.67)\n",
            "epoch: 18932, dloss = 7.6, gloss = -3144.8, total_loss = -310.7 (0.69)\n",
            "epoch: 18933, dloss = 6.5, gloss = -3143.3, total_loss = -311.1 (0.83)\n",
            "epoch: 18934, dloss = 8.8, gloss = -3136.3, total_loss = -309.2 (0.88)\n",
            "epoch: 18935, dloss = 6.8, gloss = -3117.4, total_loss = -308.3 (0.68)\n",
            "epoch: 18936, dloss = 6.2, gloss = -3139.9, total_loss = -310.9 (0.69)\n",
            "epoch: 18937, dloss = 7.3, gloss = -3137.7, total_loss = -310.1 (0.68)\n",
            "epoch: 18938, dloss = 6.9, gloss = -3142.9, total_loss = -310.8 (0.69)\n",
            "epoch: 18939, dloss = 8.4, gloss = -3168.6, total_loss = -312.7 (0.69)\n",
            "epoch: 18940, dloss = 6.6, gloss = -3166.2, total_loss = -313.3 (0.67)\n",
            "epoch: 18941, dloss = 4.5, gloss = -3161.5, total_loss = -313.9 (0.68)\n",
            "epoch: 18942, dloss = 6.9, gloss = -3157.0, total_loss = -312.2 (0.70)\n",
            "epoch: 18943, dloss = 6.2, gloss = -3134.5, total_loss = -310.4 (0.67)\n",
            "epoch: 18944, dloss = 9.1, gloss = -3152.7, total_loss = -310.7 (0.69)\n",
            "epoch: 18945, dloss = 7.7, gloss = -3162.4, total_loss = -312.4 (0.69)\n",
            "epoch: 18946, dloss = 6.7, gloss = -3173.8, total_loss = -314.1 (0.69)\n",
            "epoch: 18947, dloss = 5.5, gloss = -3167.1, total_loss = -314.0 (0.68)\n",
            "epoch: 18948, dloss = 7.2, gloss = -3169.7, total_loss = -313.4 (0.69)\n",
            "epoch: 18949, dloss = 6.4, gloss = -3172.6, total_loss = -314.1 (0.95)\n",
            "epoch: 18950, dloss = 7.9, gloss = -3146.6, total_loss = -310.7 (0.74)\n",
            "epoch: 18951, dloss = 6.0, gloss = -3122.5, total_loss = -309.2 (0.68)\n",
            "epoch: 18952, dloss = 19.9, gloss = -3147.4, total_loss = -304.8 (0.69)\n",
            "epoch: 18953, dloss = 5.8, gloss = -3134.0, total_loss = -310.5 (0.67)\n",
            "epoch: 18954, dloss = 13.8, gloss = -3189.1, total_loss = -312.0 (0.69)\n",
            "epoch: 18955, dloss = 15.6, gloss = -3148.2, total_loss = -307.0 (0.67)\n",
            "epoch: 18956, dloss = 6.4, gloss = -3114.2, total_loss = -308.2 (0.65)\n",
            "epoch: 18957, dloss = 4.1, gloss = -3130.3, total_loss = -311.0 (0.70)\n",
            "epoch: 18958, dloss = 5.2, gloss = -3146.4, total_loss = -312.1 (0.68)\n",
            "epoch: 18959, dloss = 4.2, gloss = -3124.4, total_loss = -310.3 (0.71)\n",
            "epoch: 18960, dloss = 3.5, gloss = -3126.7, total_loss = -310.9 (0.66)\n",
            "epoch: 18961, dloss = 8.9, gloss = -3126.1, total_loss = -308.2 (0.68)\n",
            "epoch: 18962, dloss = 21.4, gloss = -3132.2, total_loss = -302.5 (0.69)\n",
            "epoch: 18963, dloss = 7.6, gloss = -3136.7, total_loss = -309.9 (0.68)\n",
            "epoch: 18964, dloss = 7.9, gloss = -3128.6, total_loss = -308.9 (0.76)\n",
            "epoch: 18965, dloss = 12.6, gloss = -3165.7, total_loss = -310.3 (0.99)\n",
            "epoch: 18966, dloss = 6.7, gloss = -3147.8, total_loss = -311.4 (0.68)\n",
            "epoch: 18967, dloss = 10.5, gloss = -3138.3, total_loss = -308.6 (0.69)\n",
            "epoch: 18968, dloss = 6.5, gloss = -3140.9, total_loss = -310.8 (0.69)\n",
            "epoch: 18969, dloss = 7.0, gloss = -3139.1, total_loss = -310.4 (0.72)\n",
            "epoch: 18970, dloss = 7.1, gloss = -3122.4, total_loss = -308.7 (0.67)\n",
            "epoch: 18971, dloss = 7.3, gloss = -3113.4, total_loss = -307.7 (0.71)\n",
            "epoch: 18972, dloss = 6.6, gloss = -3099.5, total_loss = -306.6 (0.69)\n",
            "epoch: 18973, dloss = 7.7, gloss = -3086.0, total_loss = -304.8 (0.67)\n",
            "epoch: 18974, dloss = 6.8, gloss = -3097.1, total_loss = -306.3 (0.70)\n",
            "epoch: 18975, dloss = 5.9, gloss = -3108.3, total_loss = -307.9 (0.68)\n",
            "epoch: 18976, dloss = 6.4, gloss = -3095.9, total_loss = -306.4 (0.69)\n",
            "epoch: 18977, dloss = 7.6, gloss = -3136.5, total_loss = -309.8 (0.68)\n",
            "epoch: 18978, dloss = 5.7, gloss = -3129.0, total_loss = -310.0 (0.68)\n",
            "epoch: 18979, dloss = 7.0, gloss = -3108.0, total_loss = -307.3 (0.68)\n",
            "epoch: 18980, dloss = 7.0, gloss = -3110.5, total_loss = -307.6 (0.91)\n",
            "epoch: 18981, dloss = 11.3, gloss = -3107.1, total_loss = -305.1 (0.80)\n",
            "epoch: 18982, dloss = 6.4, gloss = -3125.1, total_loss = -309.3 (0.68)\n",
            "epoch: 18983, dloss = 3.5, gloss = -3141.9, total_loss = -312.4 (0.67)\n",
            "epoch: 18984, dloss = 5.5, gloss = -3132.3, total_loss = -310.5 (0.70)\n",
            "epoch: 18985, dloss = 7.1, gloss = -3134.0, total_loss = -309.8 (0.69)\n",
            "epoch: 18986, dloss = 6.6, gloss = -3115.7, total_loss = -308.3 (0.69)\n",
            "epoch: 18987, dloss = 6.8, gloss = -3111.4, total_loss = -307.7 (0.69)\n",
            "epoch: 18988, dloss = 7.2, gloss = -3126.2, total_loss = -309.0 (0.70)\n",
            "epoch: 18989, dloss = 14.9, gloss = -3123.3, total_loss = -304.9 (0.69)\n",
            "epoch: 18990, dloss = 7.0, gloss = -3144.6, total_loss = -311.0 (0.69)\n",
            "epoch: 18991, dloss = 6.1, gloss = -3134.9, total_loss = -310.4 (0.68)\n",
            "epoch: 18992, dloss = 4.6, gloss = -3126.0, total_loss = -310.3 (0.67)\n",
            "epoch: 18993, dloss = 5.2, gloss = -3140.3, total_loss = -311.4 (0.67)\n",
            "epoch: 18994, dloss = 4.5, gloss = -3131.6, total_loss = -310.9 (0.69)\n",
            "epoch: 18995, dloss = 6.2, gloss = -3145.7, total_loss = -311.5 (0.66)\n",
            "epoch: 18996, dloss = 26.5, gloss = -3153.7, total_loss = -302.1 (1.04)\n",
            "epoch: 18997, dloss = 6.9, gloss = -3125.7, total_loss = -309.1 (0.69)\n",
            "epoch: 18998, dloss = 9.5, gloss = -3134.3, total_loss = -308.7 (0.70)\n",
            "epoch: 18999, dloss = 6.9, gloss = -3160.9, total_loss = -312.6 (0.69)\n",
            "epoch: 19000, dloss = 5.9, gloss = -3120.5, total_loss = -309.1 (0.70)\n",
            "epoch: 19001, dloss = 5.1, gloss = -3142.9, total_loss = -311.7 (0.69)\n",
            "epoch: 19002, dloss = 9.7, gloss = -3155.7, total_loss = -310.7 (0.67)\n",
            "epoch: 19003, dloss = 6.5, gloss = -3128.5, total_loss = -309.6 (0.68)\n",
            "epoch: 19004, dloss = 5.4, gloss = -3149.8, total_loss = -312.3 (0.68)\n",
            "epoch: 19005, dloss = 7.3, gloss = -3157.1, total_loss = -312.1 (0.70)\n",
            "epoch: 19006, dloss = 5.6, gloss = -3150.8, total_loss = -312.3 (0.71)\n",
            "epoch: 19007, dloss = 6.7, gloss = -3169.9, total_loss = -313.6 (0.71)\n",
            "epoch: 19008, dloss = 6.8, gloss = -3174.4, total_loss = -314.0 (0.69)\n",
            "epoch: 19009, dloss = 9.3, gloss = -3164.9, total_loss = -311.8 (0.68)\n",
            "epoch: 19010, dloss = 6.4, gloss = -3157.0, total_loss = -312.5 (0.68)\n",
            "epoch: 19011, dloss = 6.4, gloss = -3148.1, total_loss = -311.6 (0.90)\n",
            "epoch: 19012, dloss = 6.6, gloss = -3135.5, total_loss = -310.2 (0.85)\n",
            "epoch: 19013, dloss = 7.7, gloss = -3155.5, total_loss = -311.7 (0.68)\n",
            "epoch: 19014, dloss = 6.7, gloss = -3126.9, total_loss = -309.4 (0.70)\n",
            "epoch: 19015, dloss = 5.6, gloss = -3147.8, total_loss = -312.0 (0.69)\n",
            "epoch: 19016, dloss = 13.8, gloss = -3152.9, total_loss = -308.4 (0.67)\n",
            "epoch: 19017, dloss = 7.6, gloss = -3145.3, total_loss = -310.7 (0.67)\n",
            "epoch: 19018, dloss = 8.4, gloss = -3143.0, total_loss = -310.1 (0.68)\n",
            "epoch: 19019, dloss = 6.4, gloss = -3124.3, total_loss = -309.2 (0.82)\n",
            "epoch: 19020, dloss = 18.7, gloss = -3168.4, total_loss = -307.5 (1.00)\n",
            "epoch: 19021, dloss = 7.9, gloss = -3148.6, total_loss = -310.9 (0.68)\n",
            "epoch: 19022, dloss = 5.8, gloss = -3160.9, total_loss = -313.2 (0.67)\n",
            "epoch: 19023, dloss = 5.3, gloss = -3180.1, total_loss = -315.3 (0.68)\n",
            "epoch: 19024, dloss = 6.4, gloss = -3172.2, total_loss = -314.0 (0.68)\n",
            "epoch: 19025, dloss = 13.5, gloss = -3172.8, total_loss = -310.5 (0.70)\n",
            "epoch: 19026, dloss = 12.2, gloss = -3184.6, total_loss = -312.4 (0.89)\n",
            "epoch: 19027, dloss = 8.1, gloss = -3177.5, total_loss = -313.7 (0.82)\n",
            "epoch: 19028, dloss = 6.9, gloss = -3178.8, total_loss = -314.4 (0.70)\n",
            "epoch: 19029, dloss = 6.9, gloss = -3199.8, total_loss = -316.5 (0.71)\n",
            "epoch: 19030, dloss = 5.2, gloss = -3193.5, total_loss = -316.8 (0.67)\n",
            "epoch: 19031, dloss = 30.8, gloss = -3229.8, total_loss = -307.6 (0.67)\n",
            "epoch: 19032, dloss = 8.0, gloss = -3211.2, total_loss = -317.1 (0.69)\n",
            "epoch: 19033, dloss = 7.6, gloss = -3215.2, total_loss = -317.7 (0.68)\n",
            "epoch: 19034, dloss = 7.1, gloss = -3214.7, total_loss = -317.9 (0.70)\n",
            "epoch: 19035, dloss = 5.9, gloss = -3204.9, total_loss = -317.5 (0.68)\n",
            "epoch: 19036, dloss = 21.8, gloss = -3258.2, total_loss = -314.9 (0.65)\n",
            "epoch: 19037, dloss = 8.7, gloss = -3244.4, total_loss = -320.1 (0.71)\n",
            "epoch: 19038, dloss = 6.6, gloss = -3258.6, total_loss = -322.6 (0.68)\n",
            "epoch: 19039, dloss = 8.7, gloss = -3237.9, total_loss = -319.4 (0.69)\n",
            "epoch: 19040, dloss = 17.5, gloss = -3226.3, total_loss = -313.9 (0.67)\n",
            "epoch: 19041, dloss = 7.6, gloss = -3210.0, total_loss = -317.2 (0.70)\n",
            "epoch: 19042, dloss = 6.1, gloss = -3214.3, total_loss = -318.4 (1.02)\n",
            "epoch: 19043, dloss = 7.6, gloss = -3228.5, total_loss = -319.0 (0.68)\n",
            "epoch: 19044, dloss = 5.9, gloss = -3203.9, total_loss = -317.4 (0.72)\n",
            "epoch: 19045, dloss = 13.2, gloss = -3230.7, total_loss = -316.5 (0.69)\n",
            "epoch: 19046, dloss = 6.3, gloss = -3250.8, total_loss = -321.9 (0.70)\n",
            "epoch: 19047, dloss = 14.3, gloss = -3212.0, total_loss = -314.0 (0.70)\n",
            "epoch: 19048, dloss = 7.5, gloss = -3241.9, total_loss = -320.4 (0.69)\n",
            "epoch: 19049, dloss = 8.3, gloss = -3248.5, total_loss = -320.7 (0.66)\n",
            "epoch: 19050, dloss = 7.7, gloss = -3234.2, total_loss = -319.6 (0.67)\n",
            "epoch: 19051, dloss = 6.1, gloss = -3231.4, total_loss = -320.1 (0.70)\n",
            "epoch: 19052, dloss = 6.9, gloss = -3233.7, total_loss = -319.9 (0.68)\n",
            "epoch: 19053, dloss = 5.5, gloss = -3220.7, total_loss = -319.3 (0.69)\n",
            "epoch: 19054, dloss = 5.3, gloss = -3228.5, total_loss = -320.2 (0.68)\n",
            "epoch: 19055, dloss = 5.7, gloss = -3234.2, total_loss = -320.6 (0.69)\n",
            "epoch: 19056, dloss = 6.6, gloss = -3248.0, total_loss = -321.5 (0.65)\n",
            "epoch: 19057, dloss = 7.8, gloss = -3249.5, total_loss = -321.0 (0.81)\n",
            "epoch: 19058, dloss = 7.9, gloss = -3253.1, total_loss = -321.3 (0.89)\n",
            "epoch: 19059, dloss = 8.2, gloss = -3228.4, total_loss = -318.8 (0.68)\n",
            "epoch: 19060, dloss = 7.4, gloss = -3238.6, total_loss = -320.2 (0.66)\n",
            "epoch: 19061, dloss = 11.4, gloss = -3250.0, total_loss = -319.3 (0.68)\n",
            "epoch: 19062, dloss = 8.6, gloss = -3237.3, total_loss = -319.4 (0.70)\n",
            "epoch: 19063, dloss = 7.2, gloss = -3227.5, total_loss = -319.2 (0.68)\n",
            "epoch: 19064, dloss = 7.7, gloss = -3241.0, total_loss = -320.3 (0.68)\n",
            "epoch: 19065, dloss = 7.3, gloss = -3233.6, total_loss = -319.7 (0.67)\n",
            "epoch: 19066, dloss = 7.5, gloss = -3239.0, total_loss = -320.1 (0.66)\n",
            "epoch: 19067, dloss = 7.1, gloss = -3223.3, total_loss = -318.8 (0.68)\n",
            "epoch: 19068, dloss = 7.6, gloss = -3221.2, total_loss = -318.3 (0.68)\n",
            "epoch: 19069, dloss = 5.8, gloss = -3234.5, total_loss = -320.6 (0.66)\n",
            "epoch: 19070, dloss = 13.3, gloss = -3188.8, total_loss = -312.2 (0.69)\n",
            "epoch: 19071, dloss = 8.3, gloss = -3239.0, total_loss = -319.8 (0.69)\n",
            "epoch: 19072, dloss = 5.4, gloss = -3202.7, total_loss = -317.6 (0.69)\n",
            "epoch: 19073, dloss = 17.0, gloss = -3232.3, total_loss = -314.7 (0.94)\n",
            "epoch: 19074, dloss = 8.1, gloss = -3240.2, total_loss = -320.0 (0.78)\n",
            "epoch: 19075, dloss = 6.7, gloss = -3250.2, total_loss = -321.7 (0.68)\n",
            "epoch: 19076, dloss = 10.2, gloss = -3229.0, total_loss = -317.8 (0.68)\n",
            "epoch: 19077, dloss = 6.3, gloss = -3223.0, total_loss = -319.1 (0.69)\n",
            "epoch: 19078, dloss = 7.2, gloss = -3240.5, total_loss = -320.5 (0.68)\n",
            "epoch: 19079, dloss = 7.8, gloss = -3244.6, total_loss = -320.5 (0.67)\n",
            "epoch: 19080, dloss = 8.0, gloss = -3238.6, total_loss = -319.9 (0.71)\n",
            "epoch: 19081, dloss = 6.8, gloss = -3255.3, total_loss = -322.1 (0.71)\n",
            "epoch: 19082, dloss = 12.4, gloss = -3258.4, total_loss = -319.6 (0.69)\n",
            "epoch: 19083, dloss = 7.6, gloss = -3248.3, total_loss = -321.1 (0.68)\n",
            "epoch: 19084, dloss = 7.5, gloss = -3252.2, total_loss = -321.5 (0.68)\n",
            "epoch: 19085, dloss = 7.8, gloss = -3250.1, total_loss = -321.1 (0.68)\n",
            "epoch: 19086, dloss = 7.2, gloss = -3258.4, total_loss = -322.2 (0.70)\n",
            "epoch: 19087, dloss = 6.7, gloss = -3261.1, total_loss = -322.7 (0.67)\n",
            "epoch: 19088, dloss = 7.4, gloss = -3255.3, total_loss = -321.8 (0.70)\n",
            "epoch: 19089, dloss = 7.3, gloss = -3266.0, total_loss = -323.0 (0.98)\n",
            "epoch: 19090, dloss = 4.7, gloss = -3273.0, total_loss = -324.9 (0.74)\n",
            "epoch: 19091, dloss = 7.2, gloss = -3283.6, total_loss = -324.7 (1.02)\n",
            "epoch: 19092, dloss = 6.7, gloss = -3265.2, total_loss = -323.2 (0.69)\n",
            "epoch: 19093, dloss = 5.7, gloss = -3276.7, total_loss = -324.8 (0.68)\n",
            "epoch: 19094, dloss = 6.5, gloss = -3280.8, total_loss = -324.8 (0.69)\n",
            "epoch: 19095, dloss = 6.2, gloss = -3280.4, total_loss = -324.9 (0.69)\n",
            "epoch: 19096, dloss = 5.4, gloss = -3286.7, total_loss = -326.0 (0.65)\n",
            "epoch: 19097, dloss = 7.1, gloss = -3280.9, total_loss = -324.5 (0.67)\n",
            "epoch: 19098, dloss = 6.4, gloss = -3274.0, total_loss = -324.2 (0.67)\n",
            "epoch: 19099, dloss = 12.4, gloss = -3285.7, total_loss = -322.4 (0.69)\n",
            "epoch: 19100, dloss = 15.8, gloss = -3273.8, total_loss = -319.5 (1.07)\n",
            "epoch: 19101, dloss = 5.8, gloss = -3286.2, total_loss = -325.7 (0.67)\n",
            "epoch: 19102, dloss = 6.8, gloss = -3258.7, total_loss = -322.5 (0.70)\n",
            "epoch: 19103, dloss = 6.7, gloss = -3263.2, total_loss = -323.0 (0.83)\n",
            "epoch: 19104, dloss = 6.9, gloss = -3259.7, total_loss = -322.5 (0.85)\n",
            "epoch: 19105, dloss = 6.5, gloss = -3263.5, total_loss = -323.1 (0.68)\n",
            "epoch: 19106, dloss = 6.5, gloss = -3274.5, total_loss = -324.2 (0.69)\n",
            "epoch: 19107, dloss = 6.6, gloss = -3262.0, total_loss = -322.9 (0.67)\n",
            "epoch: 19108, dloss = 6.2, gloss = -3275.5, total_loss = -324.5 (0.69)\n",
            "epoch: 19109, dloss = 6.2, gloss = -3304.2, total_loss = -327.3 (0.67)\n",
            "epoch: 19110, dloss = 7.6, gloss = -3276.1, total_loss = -323.8 (0.67)\n",
            "epoch: 19111, dloss = 6.7, gloss = -3268.8, total_loss = -323.5 (0.67)\n",
            "epoch: 19112, dloss = 6.2, gloss = -3262.4, total_loss = -323.1 (0.68)\n",
            "epoch: 19113, dloss = 6.3, gloss = -3234.6, total_loss = -320.3 (0.68)\n",
            "epoch: 19114, dloss = 8.8, gloss = -3257.8, total_loss = -321.4 (0.69)\n",
            "epoch: 19115, dloss = 7.9, gloss = -3272.8, total_loss = -323.3 (0.69)\n",
            "epoch: 19116, dloss = 6.7, gloss = -3278.1, total_loss = -324.4 (0.69)\n",
            "epoch: 19117, dloss = 6.9, gloss = -3274.4, total_loss = -324.0 (0.70)\n",
            "epoch: 19118, dloss = 7.0, gloss = -3263.7, total_loss = -322.9 (0.70)\n",
            "epoch: 19119, dloss = 8.2, gloss = -3264.7, total_loss = -322.4 (1.01)\n",
            "epoch: 19120, dloss = 6.4, gloss = -3267.8, total_loss = -323.6 (0.72)\n",
            "epoch: 19121, dloss = 7.6, gloss = -3259.5, total_loss = -322.1 (0.68)\n",
            "epoch: 19122, dloss = 6.2, gloss = -3248.0, total_loss = -321.7 (0.69)\n",
            "epoch: 19123, dloss = 7.4, gloss = -3277.9, total_loss = -324.1 (0.68)\n",
            "epoch: 19124, dloss = 19.4, gloss = -3297.6, total_loss = -320.1 (0.69)\n",
            "epoch: 19125, dloss = 6.6, gloss = -3297.2, total_loss = -326.4 (0.69)\n",
            "epoch: 19126, dloss = 9.7, gloss = -3275.9, total_loss = -322.8 (0.69)\n",
            "epoch: 19127, dloss = 6.3, gloss = -3269.2, total_loss = -323.8 (0.69)\n",
            "epoch: 19128, dloss = 6.8, gloss = -3272.8, total_loss = -323.9 (0.68)\n",
            "epoch: 19129, dloss = 6.5, gloss = -3268.9, total_loss = -323.6 (0.69)\n",
            "epoch: 19130, dloss = 7.7, gloss = -3286.3, total_loss = -324.8 (0.70)\n",
            "epoch: 19131, dloss = 8.1, gloss = -3256.4, total_loss = -321.6 (0.67)\n",
            "epoch: 19132, dloss = 5.3, gloss = -3266.9, total_loss = -324.0 (0.67)\n",
            "epoch: 19133, dloss = 7.4, gloss = -3283.8, total_loss = -324.7 (0.71)\n",
            "epoch: 19134, dloss = 8.7, gloss = -3281.1, total_loss = -323.7 (0.83)\n",
            "epoch: 19135, dloss = 7.8, gloss = -3250.9, total_loss = -321.2 (0.94)\n",
            "epoch: 19136, dloss = 8.1, gloss = -3261.9, total_loss = -322.2 (0.70)\n",
            "epoch: 19137, dloss = 7.5, gloss = -3228.8, total_loss = -319.2 (0.68)\n",
            "epoch: 19138, dloss = 7.5, gloss = -3232.4, total_loss = -319.5 (0.69)\n",
            "epoch: 19139, dloss = 4.7, gloss = -3247.4, total_loss = -322.4 (0.66)\n",
            "epoch: 19140, dloss = 10.0, gloss = -3272.4, total_loss = -322.2 (0.65)\n",
            "epoch: 19141, dloss = 6.7, gloss = -3257.1, total_loss = -322.3 (0.68)\n",
            "epoch: 19142, dloss = 29.0, gloss = -3253.3, total_loss = -310.8 (0.67)\n",
            "epoch: 19143, dloss = 8.2, gloss = -3255.7, total_loss = -321.5 (0.69)\n",
            "epoch: 19144, dloss = 8.5, gloss = -3272.4, total_loss = -323.0 (0.68)\n",
            "epoch: 19145, dloss = 7.9, gloss = -3279.3, total_loss = -324.0 (0.67)\n",
            "epoch: 19146, dloss = 8.5, gloss = -3296.0, total_loss = -325.3 (0.68)\n",
            "epoch: 19147, dloss = 7.8, gloss = -3277.3, total_loss = -323.9 (0.69)\n",
            "epoch: 19148, dloss = 8.0, gloss = -3301.1, total_loss = -326.1 (0.67)\n",
            "epoch: 19149, dloss = 7.9, gloss = -3298.9, total_loss = -326.0 (0.68)\n",
            "epoch: 19150, dloss = 7.8, gloss = -3304.9, total_loss = -326.6 (0.88)\n",
            "epoch: 19151, dloss = 7.4, gloss = -3285.5, total_loss = -324.9 (0.83)\n",
            "epoch: 19152, dloss = 10.9, gloss = -3286.8, total_loss = -323.2 (0.69)\n",
            "epoch: 19153, dloss = 7.7, gloss = -3276.3, total_loss = -323.8 (0.68)\n",
            "epoch: 19154, dloss = 24.2, gloss = -3295.7, total_loss = -317.5 (0.69)\n",
            "epoch: 19155, dloss = 11.8, gloss = -3279.3, total_loss = -322.0 (0.67)\n",
            "epoch: 19156, dloss = 22.2, gloss = -3292.5, total_loss = -318.1 (0.69)\n",
            "epoch: 19157, dloss = 8.4, gloss = -3286.5, total_loss = -324.4 (0.67)\n",
            "epoch: 19158, dloss = 13.0, gloss = -3306.7, total_loss = -324.2 (0.68)\n",
            "epoch: 19159, dloss = 10.8, gloss = -3310.9, total_loss = -325.7 (0.67)\n",
            "epoch: 19160, dloss = 7.7, gloss = -3309.9, total_loss = -327.1 (0.69)\n",
            "epoch: 19161, dloss = 8.2, gloss = -3331.7, total_loss = -329.1 (0.71)\n",
            "epoch: 19162, dloss = 12.9, gloss = -3331.7, total_loss = -326.7 (0.67)\n",
            "epoch: 19163, dloss = 22.6, gloss = -3332.0, total_loss = -321.9 (0.67)\n",
            "epoch: 19164, dloss = 8.9, gloss = -3340.8, total_loss = -329.7 (0.69)\n",
            "epoch: 19165, dloss = 8.4, gloss = -3304.9, total_loss = -326.3 (0.67)\n",
            "epoch: 19166, dloss = 12.2, gloss = -3329.8, total_loss = -326.9 (0.98)\n",
            "epoch: 19167, dloss = 9.8, gloss = -3312.8, total_loss = -326.4 (0.73)\n",
            "epoch: 19168, dloss = 9.1, gloss = -3311.2, total_loss = -326.6 (0.68)\n",
            "epoch: 19169, dloss = 27.7, gloss = -3290.0, total_loss = -315.1 (0.69)\n",
            "epoch: 19170, dloss = 8.3, gloss = -3290.7, total_loss = -324.9 (0.70)\n",
            "epoch: 19171, dloss = 9.8, gloss = -3279.5, total_loss = -323.1 (0.69)\n",
            "epoch: 19172, dloss = 9.8, gloss = -3282.2, total_loss = -323.3 (0.67)\n",
            "epoch: 19173, dloss = 7.5, gloss = -3278.2, total_loss = -324.0 (0.69)\n",
            "epoch: 19174, dloss = 6.3, gloss = -3276.9, total_loss = -324.6 (0.69)\n",
            "epoch: 19175, dloss = 8.8, gloss = -3288.9, total_loss = -324.5 (0.66)\n",
            "epoch: 19176, dloss = 8.3, gloss = -3282.6, total_loss = -324.1 (0.70)\n",
            "epoch: 19177, dloss = 8.9, gloss = -3313.4, total_loss = -326.9 (0.68)\n",
            "epoch: 19178, dloss = 10.7, gloss = -3318.9, total_loss = -326.5 (0.69)\n",
            "epoch: 19179, dloss = 8.3, gloss = -3304.0, total_loss = -326.3 (0.68)\n",
            "epoch: 19180, dloss = 10.9, gloss = -3334.3, total_loss = -328.0 (0.67)\n",
            "epoch: 19181, dloss = 28.8, gloss = -3328.3, total_loss = -318.4 (0.77)\n",
            "epoch: 19182, dloss = 27.9, gloss = -3318.7, total_loss = -317.9 (0.93)\n",
            "epoch: 19183, dloss = 18.8, gloss = -3330.1, total_loss = -323.6 (0.70)\n",
            "epoch: 19184, dloss = 7.8, gloss = -3321.2, total_loss = -328.2 (0.70)\n",
            "epoch: 19185, dloss = 8.6, gloss = -3330.6, total_loss = -328.8 (0.67)\n",
            "epoch: 19186, dloss = 7.8, gloss = -3344.2, total_loss = -330.5 (0.69)\n",
            "epoch: 19187, dloss = 20.5, gloss = -3371.1, total_loss = -326.8 (0.69)\n",
            "epoch: 19188, dloss = 22.8, gloss = -3363.0, total_loss = -324.9 (0.68)\n",
            "epoch: 19189, dloss = 10.2, gloss = -3361.5, total_loss = -331.0 (0.69)\n",
            "epoch: 19190, dloss = 21.1, gloss = -3352.4, total_loss = -324.7 (0.67)\n",
            "epoch: 19191, dloss = 7.1, gloss = -3351.1, total_loss = -331.6 (0.68)\n",
            "epoch: 19192, dloss = 9.8, gloss = -3355.5, total_loss = -330.6 (0.68)\n",
            "epoch: 19193, dloss = 8.5, gloss = -3342.0, total_loss = -330.0 (0.67)\n",
            "epoch: 19194, dloss = 8.6, gloss = -3361.9, total_loss = -331.9 (0.67)\n",
            "epoch: 19195, dloss = 10.1, gloss = -3348.1, total_loss = -329.8 (0.67)\n",
            "epoch: 19196, dloss = 15.3, gloss = -3372.0, total_loss = -329.6 (0.69)\n",
            "epoch: 19197, dloss = 8.5, gloss = -3358.6, total_loss = -331.6 (0.88)\n",
            "epoch: 19198, dloss = 30.0, gloss = -3363.2, total_loss = -321.3 (0.85)\n",
            "epoch: 19199, dloss = 9.7, gloss = -3331.3, total_loss = -328.3 (0.67)\n",
            "epoch: 19200, dloss = 14.7, gloss = -3340.9, total_loss = -326.8 (0.68)\n",
            "epoch: 19201, dloss = 8.8, gloss = -3350.2, total_loss = -330.6 (0.68)\n",
            "epoch: 19202, dloss = 8.1, gloss = -3363.5, total_loss = -332.3 (0.67)\n",
            "epoch: 19203, dloss = 6.6, gloss = -3366.6, total_loss = -333.3 (0.68)\n",
            "epoch: 19204, dloss = 8.0, gloss = -3362.4, total_loss = -332.2 (0.68)\n",
            "epoch: 19205, dloss = 7.5, gloss = -3365.1, total_loss = -332.8 (0.67)\n",
            "epoch: 19206, dloss = 8.5, gloss = -3380.2, total_loss = -333.8 (0.69)\n",
            "epoch: 19207, dloss = 8.1, gloss = -3383.4, total_loss = -334.3 (0.67)\n",
            "epoch: 19208, dloss = 17.2, gloss = -3376.0, total_loss = -329.0 (0.65)\n",
            "epoch: 19209, dloss = 9.6, gloss = -3368.0, total_loss = -332.0 (0.68)\n",
            "epoch: 19210, dloss = 8.5, gloss = -3356.0, total_loss = -331.4 (0.67)\n",
            "epoch: 19211, dloss = 7.4, gloss = -3351.5, total_loss = -331.4 (0.66)\n",
            "epoch: 19212, dloss = 8.0, gloss = -3362.7, total_loss = -332.3 (0.69)\n",
            "epoch: 19213, dloss = 6.7, gloss = -3361.4, total_loss = -332.8 (0.96)\n",
            "epoch: 19214, dloss = 9.4, gloss = -3345.7, total_loss = -329.9 (0.77)\n",
            "epoch: 19215, dloss = 7.9, gloss = -3357.7, total_loss = -331.8 (0.70)\n",
            "epoch: 19216, dloss = 6.9, gloss = -3362.7, total_loss = -332.8 (0.69)\n",
            "epoch: 19217, dloss = 9.3, gloss = -3367.4, total_loss = -332.1 (0.67)\n",
            "epoch: 19218, dloss = 8.8, gloss = -3360.3, total_loss = -331.6 (0.69)\n",
            "epoch: 19219, dloss = 10.3, gloss = -3365.3, total_loss = -331.4 (0.68)\n",
            "epoch: 19220, dloss = 9.5, gloss = -3395.9, total_loss = -334.9 (0.69)\n",
            "epoch: 19221, dloss = 9.7, gloss = -3391.3, total_loss = -334.3 (0.68)\n",
            "epoch: 19222, dloss = 7.7, gloss = -3385.9, total_loss = -334.8 (0.69)\n",
            "epoch: 19223, dloss = 10.7, gloss = -3387.1, total_loss = -333.4 (0.68)\n",
            "epoch: 19224, dloss = 10.4, gloss = -3398.4, total_loss = -334.7 (0.70)\n",
            "epoch: 19225, dloss = 9.9, gloss = -3395.0, total_loss = -334.5 (0.67)\n",
            "epoch: 19226, dloss = 5.8, gloss = -3367.0, total_loss = -333.8 (0.66)\n",
            "epoch: 19227, dloss = 30.0, gloss = -3371.1, total_loss = -322.1 (0.71)\n",
            "epoch: 19228, dloss = 14.7, gloss = -3358.2, total_loss = -328.5 (0.77)\n",
            "epoch: 19229, dloss = 8.9, gloss = -3339.4, total_loss = -329.5 (0.97)\n",
            "epoch: 19230, dloss = 8.6, gloss = -3357.9, total_loss = -331.5 (0.69)\n",
            "epoch: 19231, dloss = 8.8, gloss = -3371.5, total_loss = -332.7 (0.68)\n",
            "epoch: 19232, dloss = 8.4, gloss = -3381.5, total_loss = -333.9 (0.70)\n",
            "epoch: 19233, dloss = 8.3, gloss = -3389.3, total_loss = -334.8 (0.68)\n",
            "epoch: 19234, dloss = 8.3, gloss = -3390.4, total_loss = -334.9 (0.68)\n",
            "epoch: 19235, dloss = 10.3, gloss = -3379.9, total_loss = -332.8 (0.68)\n",
            "epoch: 19236, dloss = 8.9, gloss = -3396.8, total_loss = -335.2 (0.67)\n",
            "epoch: 19237, dloss = 12.5, gloss = -3393.8, total_loss = -333.1 (0.68)\n",
            "epoch: 19238, dloss = 8.5, gloss = -3383.8, total_loss = -334.1 (0.68)\n",
            "epoch: 19239, dloss = 9.4, gloss = -3396.5, total_loss = -334.9 (0.65)\n",
            "epoch: 19240, dloss = 8.8, gloss = -3402.6, total_loss = -335.9 (0.69)\n",
            "epoch: 19241, dloss = 9.7, gloss = -3397.0, total_loss = -334.8 (0.69)\n",
            "epoch: 19242, dloss = 19.3, gloss = -3389.2, total_loss = -329.3 (0.70)\n",
            "epoch: 19243, dloss = 7.6, gloss = -3387.9, total_loss = -335.0 (0.69)\n",
            "epoch: 19244, dloss = 7.4, gloss = -3394.0, total_loss = -335.7 (0.89)\n",
            "epoch: 19245, dloss = 11.1, gloss = -3373.8, total_loss = -331.8 (0.85)\n",
            "epoch: 19246, dloss = 6.9, gloss = -3347.9, total_loss = -331.4 (0.70)\n",
            "epoch: 19247, dloss = 8.9, gloss = -3362.7, total_loss = -331.8 (0.69)\n",
            "epoch: 19248, dloss = 7.1, gloss = -3345.3, total_loss = -331.0 (0.69)\n",
            "epoch: 19249, dloss = 9.4, gloss = -3366.1, total_loss = -331.9 (0.69)\n",
            "epoch: 19250, dloss = 6.0, gloss = -3339.1, total_loss = -330.9 (0.69)\n",
            "epoch: 19251, dloss = 11.4, gloss = -3375.5, total_loss = -331.8 (0.67)\n",
            "epoch: 19252, dloss = 6.8, gloss = -3360.0, total_loss = -332.6 (0.70)\n",
            "epoch: 19253, dloss = 10.5, gloss = -3371.2, total_loss = -331.9 (0.69)\n",
            "epoch: 19254, dloss = 9.3, gloss = -3343.3, total_loss = -329.7 (0.68)\n",
            "epoch: 19255, dloss = 10.0, gloss = -3376.2, total_loss = -332.6 (0.67)\n",
            "epoch: 19256, dloss = 5.2, gloss = -3359.7, total_loss = -333.3 (0.68)\n",
            "epoch: 19257, dloss = 9.2, gloss = -3377.8, total_loss = -333.2 (0.68)\n",
            "epoch: 19258, dloss = 9.6, gloss = -3376.7, total_loss = -332.9 (0.71)\n",
            "epoch: 19259, dloss = 9.9, gloss = -3366.8, total_loss = -331.7 (0.68)\n",
            "epoch: 19260, dloss = 9.5, gloss = -3363.7, total_loss = -331.6 (1.01)\n",
            "epoch: 19261, dloss = 9.5, gloss = -3371.9, total_loss = -332.4 (0.68)\n",
            "epoch: 19262, dloss = 10.0, gloss = -3398.0, total_loss = -334.8 (0.67)\n",
            "epoch: 19263, dloss = 9.4, gloss = -3394.5, total_loss = -334.7 (0.70)\n",
            "epoch: 19264, dloss = 9.1, gloss = -3383.5, total_loss = -333.8 (0.68)\n",
            "epoch: 19265, dloss = 8.6, gloss = -3386.9, total_loss = -334.4 (0.67)\n",
            "epoch: 19266, dloss = 9.0, gloss = -3402.3, total_loss = -335.7 (0.70)\n",
            "epoch: 19267, dloss = 9.3, gloss = -3396.2, total_loss = -335.0 (0.69)\n",
            "epoch: 19268, dloss = 9.3, gloss = -3397.7, total_loss = -335.1 (0.66)\n",
            "epoch: 19269, dloss = 9.7, gloss = -3392.2, total_loss = -334.4 (0.66)\n",
            "epoch: 19270, dloss = 10.9, gloss = -3379.8, total_loss = -332.6 (0.65)\n",
            "epoch: 19271, dloss = 9.3, gloss = -3378.6, total_loss = -333.2 (0.67)\n",
            "epoch: 19272, dloss = 7.7, gloss = -3380.1, total_loss = -334.1 (0.68)\n",
            "epoch: 19273, dloss = 10.3, gloss = -3379.9, total_loss = -332.8 (0.68)\n",
            "epoch: 19274, dloss = 7.1, gloss = -3381.7, total_loss = -334.6 (0.68)\n",
            "epoch: 19275, dloss = 8.5, gloss = -3383.9, total_loss = -334.1 (0.75)\n",
            "epoch: 19276, dloss = 10.1, gloss = -3395.8, total_loss = -334.5 (0.97)\n",
            "epoch: 19277, dloss = 9.8, gloss = -3389.8, total_loss = -334.1 (0.68)\n",
            "epoch: 19278, dloss = 6.9, gloss = -3401.9, total_loss = -336.8 (0.69)\n",
            "epoch: 19279, dloss = 10.1, gloss = -3387.2, total_loss = -333.7 (0.67)\n",
            "epoch: 19280, dloss = 8.7, gloss = -3402.0, total_loss = -335.9 (0.66)\n",
            "epoch: 19281, dloss = 9.1, gloss = -3410.2, total_loss = -336.5 (0.66)\n",
            "epoch: 19282, dloss = 7.3, gloss = -3396.8, total_loss = -336.0 (0.68)\n",
            "epoch: 19283, dloss = 9.1, gloss = -3402.0, total_loss = -335.6 (0.66)\n",
            "epoch: 19284, dloss = 9.1, gloss = -3396.5, total_loss = -335.1 (0.65)\n",
            "epoch: 19285, dloss = 8.0, gloss = -3390.3, total_loss = -335.0 (0.68)\n",
            "epoch: 19286, dloss = 9.3, gloss = -3404.6, total_loss = -335.8 (0.68)\n",
            "epoch: 19287, dloss = 8.3, gloss = -3397.4, total_loss = -335.6 (0.68)\n",
            "epoch: 19288, dloss = 9.2, gloss = -3369.0, total_loss = -332.3 (0.68)\n",
            "epoch: 19289, dloss = 9.1, gloss = -3375.1, total_loss = -332.9 (0.69)\n",
            "epoch: 19290, dloss = 9.6, gloss = -3389.9, total_loss = -334.2 (0.68)\n",
            "epoch: 19291, dloss = 10.0, gloss = -3395.4, total_loss = -334.5 (0.82)\n",
            "epoch: 19292, dloss = 10.6, gloss = -3426.8, total_loss = -337.4 (0.90)\n",
            "epoch: 19293, dloss = 9.7, gloss = -3412.8, total_loss = -336.4 (0.68)\n",
            "epoch: 19294, dloss = 11.4, gloss = -3394.8, total_loss = -333.8 (0.68)\n",
            "epoch: 19295, dloss = 12.0, gloss = -3392.9, total_loss = -333.3 (0.69)\n",
            "epoch: 19296, dloss = 8.2, gloss = -3387.3, total_loss = -334.6 (0.68)\n",
            "epoch: 19297, dloss = 8.9, gloss = -3407.0, total_loss = -336.2 (0.68)\n",
            "epoch: 19298, dloss = 9.3, gloss = -3404.6, total_loss = -335.8 (0.70)\n",
            "epoch: 19299, dloss = 17.7, gloss = -3385.6, total_loss = -329.7 (0.68)\n",
            "epoch: 19300, dloss = 9.9, gloss = -3383.6, total_loss = -333.4 (0.67)\n",
            "epoch: 19301, dloss = 13.2, gloss = -3417.2, total_loss = -335.1 (0.69)\n",
            "epoch: 19302, dloss = 9.3, gloss = -3408.5, total_loss = -336.2 (0.69)\n",
            "epoch: 19303, dloss = 7.3, gloss = -3412.7, total_loss = -337.6 (0.69)\n",
            "epoch: 19304, dloss = 9.1, gloss = -3427.4, total_loss = -338.2 (0.69)\n",
            "epoch: 19305, dloss = 8.7, gloss = -3405.6, total_loss = -336.2 (0.69)\n",
            "epoch: 19306, dloss = 11.0, gloss = -3418.8, total_loss = -336.4 (0.70)\n",
            "epoch: 19307, dloss = 9.0, gloss = -3425.8, total_loss = -338.1 (0.96)\n",
            "epoch: 19308, dloss = 9.3, gloss = -3409.8, total_loss = -336.3 (0.76)\n",
            "epoch: 19309, dloss = 8.3, gloss = -3419.9, total_loss = -337.8 (0.67)\n",
            "epoch: 19310, dloss = 7.8, gloss = -3426.3, total_loss = -338.8 (0.69)\n",
            "epoch: 19311, dloss = 7.8, gloss = -3417.5, total_loss = -337.9 (0.68)\n",
            "epoch: 19312, dloss = 7.8, gloss = -3426.4, total_loss = -338.7 (0.68)\n",
            "epoch: 19313, dloss = 8.5, gloss = -3415.9, total_loss = -337.4 (0.70)\n",
            "epoch: 19314, dloss = 8.7, gloss = -3409.5, total_loss = -336.6 (0.68)\n",
            "epoch: 19315, dloss = 9.3, gloss = -3399.5, total_loss = -335.3 (0.69)\n",
            "epoch: 19316, dloss = 12.0, gloss = -3399.4, total_loss = -333.9 (0.70)\n",
            "epoch: 19317, dloss = 9.1, gloss = -3416.4, total_loss = -337.1 (0.86)\n",
            "epoch: 19318, dloss = 9.8, gloss = -3430.4, total_loss = -338.2 (0.99)\n",
            "epoch: 19319, dloss = 8.7, gloss = -3419.6, total_loss = -337.6 (0.69)\n",
            "epoch: 19320, dloss = 9.6, gloss = -3415.5, total_loss = -336.7 (0.68)\n",
            "epoch: 19321, dloss = 9.8, gloss = -3416.1, total_loss = -336.7 (0.68)\n",
            "epoch: 19322, dloss = 8.7, gloss = -3400.4, total_loss = -335.7 (1.06)\n",
            "epoch: 19323, dloss = 10.2, gloss = -3409.8, total_loss = -335.9 (0.69)\n",
            "epoch: 19324, dloss = 14.5, gloss = -3405.2, total_loss = -333.3 (0.69)\n",
            "epoch: 19325, dloss = 10.2, gloss = -3402.1, total_loss = -335.1 (0.68)\n",
            "epoch: 19326, dloss = 9.2, gloss = -3405.1, total_loss = -335.9 (0.68)\n",
            "epoch: 19327, dloss = 10.4, gloss = -3403.0, total_loss = -335.1 (0.68)\n",
            "epoch: 19328, dloss = 8.2, gloss = -3421.0, total_loss = -338.0 (0.67)\n",
            "epoch: 19329, dloss = 13.9, gloss = -3406.0, total_loss = -333.6 (0.69)\n",
            "epoch: 19330, dloss = 11.8, gloss = -3401.3, total_loss = -334.2 (0.69)\n",
            "epoch: 19331, dloss = 10.2, gloss = -3406.7, total_loss = -335.6 (0.65)\n",
            "epoch: 19332, dloss = 10.5, gloss = -3422.4, total_loss = -337.0 (0.66)\n",
            "epoch: 19333, dloss = 11.9, gloss = -3433.9, total_loss = -337.4 (0.69)\n",
            "epoch: 19334, dloss = 9.9, gloss = -3446.7, total_loss = -339.7 (0.69)\n",
            "epoch: 19335, dloss = 24.6, gloss = -3417.3, total_loss = -329.4 (0.69)\n",
            "epoch: 19336, dloss = 10.5, gloss = -3431.0, total_loss = -337.8 (0.70)\n",
            "epoch: 19337, dloss = 12.1, gloss = -3426.9, total_loss = -336.6 (0.84)\n",
            "epoch: 19338, dloss = 10.6, gloss = -3442.0, total_loss = -338.9 (0.88)\n",
            "epoch: 19339, dloss = 30.9, gloss = -3431.8, total_loss = -327.7 (0.71)\n",
            "epoch: 19340, dloss = 11.5, gloss = -3429.7, total_loss = -337.2 (0.68)\n",
            "epoch: 19341, dloss = 14.7, gloss = -3433.2, total_loss = -336.0 (0.69)\n",
            "epoch: 19342, dloss = 10.4, gloss = -3428.4, total_loss = -337.6 (0.69)\n",
            "epoch: 19343, dloss = 12.8, gloss = -3434.4, total_loss = -337.1 (0.69)\n",
            "epoch: 19344, dloss = 9.5, gloss = -3445.6, total_loss = -339.8 (0.67)\n",
            "epoch: 19345, dloss = 13.9, gloss = -3451.1, total_loss = -338.1 (0.67)\n",
            "epoch: 19346, dloss = 10.1, gloss = -3461.4, total_loss = -341.1 (0.70)\n",
            "epoch: 19347, dloss = 9.8, gloss = -3459.4, total_loss = -341.0 (0.84)\n",
            "epoch: 19348, dloss = 9.2, gloss = -3457.5, total_loss = -341.2 (0.91)\n",
            "epoch: 19349, dloss = 10.6, gloss = -3480.0, total_loss = -342.7 (0.68)\n",
            "epoch: 19350, dloss = 10.9, gloss = -3466.4, total_loss = -341.2 (0.71)\n",
            "epoch: 19351, dloss = 9.1, gloss = -3463.8, total_loss = -341.8 (0.69)\n",
            "epoch: 19352, dloss = 8.6, gloss = -3460.4, total_loss = -341.8 (0.86)\n",
            "epoch: 19353, dloss = 11.9, gloss = -3497.9, total_loss = -343.8 (0.90)\n",
            "epoch: 19354, dloss = 11.5, gloss = -3473.9, total_loss = -341.7 (0.67)\n",
            "epoch: 19355, dloss = 8.6, gloss = -3473.4, total_loss = -343.1 (0.78)\n",
            "epoch: 19356, dloss = 11.0, gloss = -3475.8, total_loss = -342.1 (0.96)\n",
            "epoch: 19357, dloss = 9.7, gloss = -3470.2, total_loss = -342.2 (0.66)\n",
            "epoch: 19358, dloss = 10.1, gloss = -3475.2, total_loss = -342.5 (0.69)\n",
            "epoch: 19359, dloss = 27.4, gloss = -3468.9, total_loss = -333.2 (0.70)\n",
            "epoch: 19360, dloss = 12.1, gloss = -3434.8, total_loss = -337.4 (0.68)\n",
            "epoch: 19361, dloss = 9.8, gloss = -3451.7, total_loss = -340.3 (0.69)\n",
            "epoch: 19362, dloss = 10.2, gloss = -3439.7, total_loss = -338.9 (0.68)\n",
            "epoch: 19363, dloss = 10.1, gloss = -3451.6, total_loss = -340.1 (0.68)\n",
            "epoch: 19364, dloss = 9.7, gloss = -3463.4, total_loss = -341.5 (0.68)\n",
            "epoch: 19365, dloss = 9.0, gloss = -3476.1, total_loss = -343.1 (0.68)\n",
            "epoch: 19366, dloss = 9.3, gloss = -3462.5, total_loss = -341.6 (0.70)\n",
            "epoch: 19367, dloss = 12.0, gloss = -3473.1, total_loss = -341.3 (0.86)\n",
            "epoch: 19368, dloss = 8.7, gloss = -3446.6, total_loss = -340.3 (0.89)\n",
            "epoch: 19369, dloss = 10.7, gloss = -3460.3, total_loss = -340.7 (0.68)\n",
            "epoch: 19370, dloss = 10.3, gloss = -3442.8, total_loss = -339.1 (0.68)\n",
            "epoch: 19371, dloss = 9.4, gloss = -3448.9, total_loss = -340.2 (0.68)\n",
            "epoch: 19372, dloss = 10.7, gloss = -3468.5, total_loss = -341.5 (0.67)\n",
            "epoch: 19373, dloss = 10.5, gloss = -3481.3, total_loss = -342.9 (0.68)\n",
            "epoch: 19374, dloss = 11.6, gloss = -3466.1, total_loss = -340.8 (0.68)\n",
            "epoch: 19375, dloss = 13.2, gloss = -3461.0, total_loss = -339.5 (0.68)\n",
            "epoch: 19376, dloss = 10.8, gloss = -3471.7, total_loss = -341.8 (0.66)\n",
            "epoch: 19377, dloss = 10.6, gloss = -3478.7, total_loss = -342.6 (0.68)\n",
            "epoch: 19378, dloss = 12.0, gloss = -3477.4, total_loss = -341.8 (0.68)\n",
            "epoch: 19379, dloss = 14.0, gloss = -3479.1, total_loss = -340.9 (0.68)\n",
            "epoch: 19380, dloss = 10.0, gloss = -3464.3, total_loss = -341.4 (0.68)\n",
            "epoch: 19381, dloss = 7.9, gloss = -3462.7, total_loss = -342.3 (0.68)\n",
            "epoch: 19382, dloss = 10.7, gloss = -3488.1, total_loss = -343.4 (0.67)\n",
            "epoch: 19383, dloss = 11.2, gloss = -3492.8, total_loss = -343.7 (0.98)\n",
            "epoch: 19384, dloss = 10.1, gloss = -3483.6, total_loss = -343.3 (0.81)\n",
            "epoch: 19385, dloss = 8.6, gloss = -3479.2, total_loss = -343.6 (0.69)\n",
            "epoch: 19386, dloss = 9.7, gloss = -3487.8, total_loss = -343.9 (0.69)\n",
            "epoch: 19387, dloss = 44.3, gloss = -3490.7, total_loss = -326.9 (0.68)\n",
            "epoch: 19388, dloss = 15.6, gloss = -3483.0, total_loss = -340.5 (0.69)\n",
            "epoch: 19389, dloss = 11.5, gloss = -3461.3, total_loss = -340.4 (0.69)\n",
            "epoch: 19390, dloss = 11.9, gloss = -3491.4, total_loss = -343.2 (0.67)\n",
            "epoch: 19391, dloss = 10.0, gloss = -3479.6, total_loss = -343.0 (0.67)\n",
            "epoch: 19392, dloss = 10.7, gloss = -3472.3, total_loss = -341.9 (0.67)\n",
            "epoch: 19393, dloss = 9.7, gloss = -3469.5, total_loss = -342.1 (0.68)\n",
            "epoch: 19394, dloss = 9.7, gloss = -3477.9, total_loss = -342.9 (0.67)\n",
            "epoch: 19395, dloss = 11.2, gloss = -3476.7, total_loss = -342.1 (0.68)\n",
            "epoch: 19396, dloss = 7.9, gloss = -3493.2, total_loss = -345.4 (0.67)\n",
            "epoch: 19397, dloss = 12.1, gloss = -3488.7, total_loss = -342.8 (0.68)\n",
            "epoch: 19398, dloss = 10.5, gloss = -3501.1, total_loss = -344.8 (0.70)\n",
            "epoch: 19399, dloss = 9.1, gloss = -3500.2, total_loss = -345.5 (0.99)\n",
            "epoch: 19400, dloss = 12.3, gloss = -3490.2, total_loss = -342.8 (0.66)\n",
            "epoch: 19401, dloss = 11.3, gloss = -3490.4, total_loss = -343.4 (0.69)\n",
            "epoch: 19402, dloss = 10.2, gloss = -3487.2, total_loss = -343.6 (0.69)\n",
            "epoch: 19403, dloss = 9.4, gloss = -3509.3, total_loss = -346.2 (0.68)\n",
            "epoch: 19404, dloss = 11.1, gloss = -3514.1, total_loss = -345.9 (0.70)\n",
            "epoch: 19405, dloss = 9.1, gloss = -3503.4, total_loss = -345.8 (0.68)\n",
            "epoch: 19406, dloss = 10.1, gloss = -3490.0, total_loss = -343.9 (0.65)\n",
            "epoch: 19407, dloss = 9.4, gloss = -3489.4, total_loss = -344.2 (0.68)\n",
            "epoch: 19408, dloss = 11.5, gloss = -3468.4, total_loss = -341.1 (0.69)\n",
            "epoch: 19409, dloss = 10.7, gloss = -3484.5, total_loss = -343.1 (0.70)\n",
            "epoch: 19410, dloss = 11.1, gloss = -3477.9, total_loss = -342.3 (0.69)\n",
            "epoch: 19411, dloss = 11.2, gloss = -3470.7, total_loss = -341.5 (0.69)\n",
            "epoch: 19412, dloss = 10.5, gloss = -3485.1, total_loss = -343.3 (0.67)\n",
            "epoch: 19413, dloss = 8.7, gloss = -3490.9, total_loss = -344.7 (0.70)\n",
            "epoch: 19414, dloss = 11.5, gloss = -3474.0, total_loss = -341.6 (0.86)\n",
            "epoch: 19415, dloss = 12.1, gloss = -3474.8, total_loss = -341.4 (0.87)\n",
            "epoch: 19416, dloss = 9.9, gloss = -3492.2, total_loss = -344.3 (0.67)\n",
            "epoch: 19417, dloss = 11.6, gloss = -3481.8, total_loss = -342.4 (0.68)\n",
            "epoch: 19418, dloss = 10.0, gloss = -3484.6, total_loss = -343.5 (0.69)\n",
            "epoch: 19419, dloss = 12.8, gloss = -3473.9, total_loss = -341.0 (0.67)\n",
            "epoch: 19420, dloss = 9.6, gloss = -3471.3, total_loss = -342.3 (0.69)\n",
            "epoch: 19421, dloss = 10.7, gloss = -3473.7, total_loss = -342.0 (0.67)\n",
            "epoch: 19422, dloss = 9.1, gloss = -3476.3, total_loss = -343.1 (0.68)\n",
            "epoch: 19423, dloss = 11.7, gloss = -3483.1, total_loss = -342.4 (0.69)\n",
            "epoch: 19424, dloss = 10.3, gloss = -3456.0, total_loss = -340.4 (0.67)\n",
            "epoch: 19425, dloss = 9.2, gloss = -3453.1, total_loss = -340.7 (0.69)\n",
            "epoch: 19426, dloss = 11.6, gloss = -3489.7, total_loss = -343.2 (0.69)\n",
            "epoch: 19427, dloss = 10.2, gloss = -3484.4, total_loss = -343.3 (0.68)\n",
            "epoch: 19428, dloss = 12.0, gloss = -3480.7, total_loss = -342.1 (0.65)\n",
            "epoch: 19429, dloss = 10.5, gloss = -3480.4, total_loss = -342.8 (0.68)\n",
            "epoch: 19430, dloss = 11.2, gloss = -3490.0, total_loss = -343.4 (0.93)\n",
            "epoch: 19431, dloss = 10.9, gloss = -3487.0, total_loss = -343.2 (0.80)\n",
            "epoch: 19432, dloss = 12.8, gloss = -3478.9, total_loss = -341.5 (0.65)\n",
            "epoch: 19433, dloss = 9.7, gloss = -3493.3, total_loss = -344.5 (0.66)\n",
            "epoch: 19434, dloss = 10.9, gloss = -3482.3, total_loss = -342.8 (0.68)\n",
            "epoch: 19435, dloss = 9.7, gloss = -3488.0, total_loss = -344.0 (0.68)\n",
            "epoch: 19436, dloss = 10.0, gloss = -3489.5, total_loss = -343.9 (0.68)\n",
            "epoch: 19437, dloss = 8.5, gloss = -3498.0, total_loss = -345.5 (0.69)\n",
            "epoch: 19438, dloss = 10.5, gloss = -3494.4, total_loss = -344.2 (0.66)\n",
            "epoch: 19439, dloss = 9.0, gloss = -3493.2, total_loss = -344.8 (0.68)\n",
            "epoch: 19440, dloss = 11.9, gloss = -3502.2, total_loss = -344.3 (0.71)\n",
            "epoch: 19441, dloss = 10.5, gloss = -3517.0, total_loss = -346.4 (0.71)\n",
            "epoch: 19442, dloss = 11.8, gloss = -3507.4, total_loss = -344.9 (0.69)\n",
            "epoch: 19443, dloss = 8.9, gloss = -3494.3, total_loss = -345.0 (0.69)\n",
            "epoch: 19444, dloss = 10.0, gloss = -3483.3, total_loss = -343.3 (0.69)\n",
            "epoch: 19445, dloss = 9.7, gloss = -3479.1, total_loss = -343.0 (0.73)\n",
            "epoch: 19446, dloss = 9.1, gloss = -3491.8, total_loss = -344.6 (1.02)\n",
            "epoch: 19447, dloss = 9.9, gloss = -3503.8, total_loss = -345.4 (0.67)\n",
            "epoch: 19448, dloss = 8.4, gloss = -3488.1, total_loss = -344.6 (0.68)\n",
            "epoch: 19449, dloss = 10.4, gloss = -3483.2, total_loss = -343.1 (0.68)\n",
            "epoch: 19450, dloss = 10.8, gloss = -3494.0, total_loss = -344.0 (0.68)\n",
            "epoch: 19451, dloss = 10.8, gloss = -3492.6, total_loss = -343.8 (0.69)\n",
            "epoch: 19452, dloss = 8.4, gloss = -3494.1, total_loss = -345.2 (0.69)\n",
            "epoch: 19453, dloss = 10.2, gloss = -3496.5, total_loss = -344.5 (0.70)\n",
            "epoch: 19454, dloss = 10.9, gloss = -3486.2, total_loss = -343.2 (0.67)\n",
            "epoch: 19455, dloss = 10.3, gloss = -3485.8, total_loss = -343.4 (0.68)\n",
            "epoch: 19456, dloss = 10.2, gloss = -3486.2, total_loss = -343.5 (0.68)\n",
            "epoch: 19457, dloss = 15.1, gloss = -3482.4, total_loss = -340.7 (0.68)\n",
            "epoch: 19458, dloss = 11.6, gloss = -3492.5, total_loss = -343.4 (0.69)\n",
            "epoch: 19459, dloss = 10.6, gloss = -3502.4, total_loss = -344.9 (0.69)\n",
            "epoch: 19460, dloss = 8.4, gloss = -3504.0, total_loss = -346.2 (0.69)\n",
            "epoch: 19461, dloss = 15.0, gloss = -3505.4, total_loss = -343.0 (0.86)\n",
            "epoch: 19462, dloss = 9.0, gloss = -3478.4, total_loss = -343.4 (0.87)\n",
            "epoch: 19463, dloss = 9.9, gloss = -3506.5, total_loss = -345.7 (0.70)\n",
            "epoch: 19464, dloss = 12.2, gloss = -3503.5, total_loss = -344.2 (0.66)\n",
            "epoch: 19465, dloss = 8.0, gloss = -3499.2, total_loss = -345.9 (0.66)\n",
            "epoch: 19466, dloss = 11.5, gloss = -3505.1, total_loss = -344.8 (0.67)\n",
            "epoch: 19467, dloss = 9.7, gloss = -3527.2, total_loss = -347.9 (0.69)\n",
            "epoch: 19468, dloss = 71.2, gloss = -3524.5, total_loss = -316.8 (0.70)\n",
            "epoch: 19469, dloss = 10.7, gloss = -3512.0, total_loss = -345.9 (0.70)\n",
            "epoch: 19470, dloss = 11.6, gloss = -3504.6, total_loss = -344.7 (0.68)\n",
            "epoch: 19471, dloss = 15.5, gloss = -3499.4, total_loss = -342.2 (0.71)\n",
            "epoch: 19472, dloss = 60.5, gloss = -3525.7, total_loss = -322.3 (0.68)\n",
            "epoch: 19473, dloss = 13.7, gloss = -3523.0, total_loss = -345.4 (0.70)\n",
            "epoch: 19474, dloss = 16.0, gloss = -3509.9, total_loss = -343.0 (0.72)\n",
            "epoch: 19475, dloss = 12.3, gloss = -3512.5, total_loss = -345.1 (0.72)\n",
            "epoch: 19476, dloss = 11.9, gloss = -3493.6, total_loss = -343.4 (0.73)\n",
            "epoch: 19477, dloss = 11.2, gloss = -3502.3, total_loss = -344.6 (1.02)\n",
            "epoch: 19478, dloss = 12.5, gloss = -3513.5, total_loss = -345.1 (0.69)\n",
            "epoch: 19479, dloss = 14.3, gloss = -3509.6, total_loss = -343.8 (0.69)\n",
            "epoch: 19480, dloss = 12.7, gloss = -3517.1, total_loss = -345.4 (0.70)\n",
            "epoch: 19481, dloss = 10.7, gloss = -3503.0, total_loss = -345.0 (0.70)\n",
            "epoch: 19482, dloss = 14.2, gloss = -3516.0, total_loss = -344.5 (0.68)\n",
            "epoch: 19483, dloss = 9.4, gloss = -3530.5, total_loss = -348.4 (0.69)\n",
            "epoch: 19484, dloss = 11.6, gloss = -3529.4, total_loss = -347.1 (0.69)\n",
            "epoch: 19485, dloss = 10.3, gloss = -3526.1, total_loss = -347.4 (0.69)\n",
            "epoch: 19486, dloss = 11.8, gloss = -3529.7, total_loss = -347.1 (0.68)\n",
            "epoch: 19487, dloss = 16.8, gloss = -3529.8, total_loss = -344.6 (0.69)\n",
            "epoch: 19488, dloss = 10.8, gloss = -3508.6, total_loss = -345.5 (0.67)\n",
            "epoch: 19489, dloss = 11.9, gloss = -3514.8, total_loss = -345.5 (0.69)\n",
            "epoch: 19490, dloss = 12.1, gloss = -3499.5, total_loss = -343.9 (0.70)\n",
            "epoch: 19491, dloss = 10.9, gloss = -3532.8, total_loss = -347.8 (0.68)\n",
            "epoch: 19492, dloss = 11.6, gloss = -3535.8, total_loss = -347.8 (0.91)\n",
            "epoch: 19493, dloss = 10.1, gloss = -3523.3, total_loss = -347.3 (0.86)\n",
            "epoch: 19494, dloss = 10.8, gloss = -3538.8, total_loss = -348.5 (0.68)\n",
            "epoch: 19495, dloss = 9.5, gloss = -3519.5, total_loss = -347.2 (0.70)\n",
            "epoch: 19496, dloss = 8.5, gloss = -3534.9, total_loss = -349.2 (0.69)\n",
            "epoch: 19497, dloss = 11.7, gloss = -3535.5, total_loss = -347.7 (0.68)\n",
            "epoch: 19498, dloss = 11.0, gloss = -3535.8, total_loss = -348.1 (0.67)\n",
            "epoch: 19499, dloss = 10.3, gloss = -3534.1, total_loss = -348.3 (0.69)\n",
            "epoch: 19500, dloss = 10.0, gloss = -3545.5, total_loss = -349.5 (0.69)\n",
            "epoch: 19501, dloss = 10.7, gloss = -3528.2, total_loss = -347.4 (0.69)\n",
            "epoch: 19502, dloss = 10.3, gloss = -3524.3, total_loss = -347.3 (0.69)\n",
            "epoch: 19503, dloss = 10.7, gloss = -3546.7, total_loss = -349.3 (0.69)\n",
            "epoch: 19504, dloss = 11.1, gloss = -3535.7, total_loss = -348.0 (0.69)\n",
            "epoch: 19505, dloss = 10.5, gloss = -3537.0, total_loss = -348.4 (0.68)\n",
            "epoch: 19506, dloss = 9.8, gloss = -3539.0, total_loss = -349.0 (0.68)\n",
            "epoch: 19507, dloss = 10.8, gloss = -3550.4, total_loss = -349.6 (0.69)\n",
            "epoch: 19508, dloss = 11.4, gloss = -3548.4, total_loss = -349.2 (1.04)\n",
            "epoch: 19509, dloss = 11.4, gloss = -3550.7, total_loss = -349.4 (0.66)\n",
            "epoch: 19510, dloss = 12.0, gloss = -3557.1, total_loss = -349.7 (0.71)\n",
            "epoch: 19511, dloss = 10.2, gloss = -3560.2, total_loss = -350.9 (0.70)\n",
            "epoch: 19512, dloss = 12.4, gloss = -3552.2, total_loss = -349.0 (0.67)\n",
            "epoch: 19513, dloss = 13.2, gloss = -3562.2, total_loss = -349.6 (0.68)\n",
            "epoch: 19514, dloss = 11.0, gloss = -3561.3, total_loss = -350.6 (0.68)\n",
            "epoch: 19515, dloss = 13.4, gloss = -3566.7, total_loss = -350.0 (0.69)\n",
            "epoch: 19516, dloss = 12.4, gloss = -3588.3, total_loss = -352.6 (0.70)\n",
            "epoch: 19517, dloss = 10.8, gloss = -3572.9, total_loss = -351.9 (0.68)\n",
            "epoch: 19518, dloss = 11.5, gloss = -3551.4, total_loss = -349.4 (0.70)\n",
            "epoch: 19519, dloss = 12.3, gloss = -3571.2, total_loss = -351.0 (0.70)\n",
            "epoch: 19520, dloss = 12.2, gloss = -3555.9, total_loss = -349.5 (0.67)\n",
            "epoch: 19521, dloss = 11.8, gloss = -3569.7, total_loss = -351.1 (0.68)\n",
            "epoch: 19522, dloss = 22.0, gloss = -3579.0, total_loss = -346.9 (0.68)\n",
            "epoch: 19523, dloss = 11.9, gloss = -3567.9, total_loss = -350.8 (0.86)\n",
            "epoch: 19524, dloss = 10.1, gloss = -3563.0, total_loss = -351.3 (0.86)\n",
            "epoch: 19525, dloss = 15.2, gloss = -3559.3, total_loss = -348.3 (0.69)\n",
            "epoch: 19526, dloss = 13.0, gloss = -3567.7, total_loss = -350.3 (0.70)\n",
            "epoch: 19527, dloss = 12.8, gloss = -3578.3, total_loss = -351.4 (0.66)\n",
            "epoch: 19528, dloss = 11.8, gloss = -3559.8, total_loss = -350.1 (0.67)\n",
            "epoch: 19529, dloss = 9.6, gloss = -3567.4, total_loss = -352.0 (0.70)\n",
            "epoch: 19530, dloss = 11.3, gloss = -3560.2, total_loss = -350.4 (0.67)\n",
            "epoch: 19531, dloss = 17.3, gloss = -3560.2, total_loss = -347.3 (0.68)\n",
            "epoch: 19532, dloss = 11.8, gloss = -3565.1, total_loss = -350.6 (0.70)\n",
            "epoch: 19533, dloss = 11.1, gloss = -3562.2, total_loss = -350.7 (0.66)\n",
            "epoch: 19534, dloss = 13.0, gloss = -3566.6, total_loss = -350.2 (0.69)\n",
            "epoch: 19535, dloss = 11.8, gloss = -3561.4, total_loss = -350.2 (0.68)\n",
            "epoch: 19536, dloss = 13.7, gloss = -3558.2, total_loss = -349.0 (0.68)\n",
            "epoch: 19537, dloss = 9.3, gloss = -3562.8, total_loss = -351.6 (0.70)\n",
            "epoch: 19538, dloss = 11.6, gloss = -3565.3, total_loss = -350.8 (0.66)\n",
            "epoch: 19539, dloss = 11.7, gloss = -3571.4, total_loss = -351.3 (0.95)\n",
            "epoch: 19540, dloss = 11.3, gloss = -3566.0, total_loss = -351.0 (0.77)\n",
            "epoch: 19541, dloss = 11.6, gloss = -3554.2, total_loss = -349.6 (0.69)\n",
            "epoch: 19542, dloss = 11.0, gloss = -3573.1, total_loss = -351.8 (0.68)\n",
            "epoch: 19543, dloss = 20.6, gloss = -3560.7, total_loss = -345.8 (0.69)\n",
            "epoch: 19544, dloss = 10.6, gloss = -3571.6, total_loss = -351.9 (0.69)\n",
            "epoch: 19545, dloss = 10.2, gloss = -3576.4, total_loss = -352.6 (0.69)\n",
            "epoch: 19546, dloss = 10.4, gloss = -3563.4, total_loss = -351.1 (0.67)\n",
            "epoch: 19547, dloss = 10.3, gloss = -3584.6, total_loss = -353.3 (0.69)\n",
            "epoch: 19548, dloss = 12.5, gloss = -3574.5, total_loss = -351.2 (0.69)\n",
            "epoch: 19549, dloss = 9.2, gloss = -3574.6, total_loss = -352.9 (0.69)\n",
            "epoch: 19550, dloss = 8.6, gloss = -3568.0, total_loss = -352.5 (0.69)\n",
            "epoch: 19551, dloss = 9.6, gloss = -3568.9, total_loss = -352.1 (0.68)\n",
            "epoch: 19552, dloss = 9.8, gloss = -3583.6, total_loss = -353.4 (0.68)\n",
            "epoch: 19553, dloss = 10.8, gloss = -3577.9, total_loss = -352.4 (0.70)\n",
            "epoch: 19554, dloss = 9.7, gloss = -3583.6, total_loss = -353.5 (0.77)\n",
            "epoch: 19555, dloss = 11.0, gloss = -3593.6, total_loss = -353.9 (0.96)\n",
            "epoch: 19556, dloss = 11.0, gloss = -3587.1, total_loss = -353.2 (0.68)\n",
            "epoch: 19557, dloss = 9.9, gloss = -3580.7, total_loss = -353.1 (0.68)\n",
            "epoch: 19558, dloss = 12.4, gloss = -3576.0, total_loss = -351.4 (0.69)\n",
            "epoch: 19559, dloss = 9.4, gloss = -3598.9, total_loss = -355.2 (0.66)\n",
            "epoch: 19560, dloss = 11.9, gloss = -3594.0, total_loss = -353.5 (0.67)\n",
            "epoch: 19561, dloss = 12.3, gloss = -3574.7, total_loss = -351.3 (0.69)\n",
            "epoch: 19562, dloss = 10.2, gloss = -3567.6, total_loss = -351.7 (0.69)\n",
            "epoch: 19563, dloss = 10.5, gloss = -3595.0, total_loss = -354.3 (0.67)\n",
            "epoch: 19564, dloss = 12.3, gloss = -3576.9, total_loss = -351.5 (0.67)\n",
            "epoch: 19565, dloss = 11.0, gloss = -3582.9, total_loss = -352.8 (0.65)\n",
            "epoch: 19566, dloss = 10.9, gloss = -3585.9, total_loss = -353.1 (0.70)\n",
            "epoch: 19567, dloss = 9.6, gloss = -3577.1, total_loss = -352.9 (0.71)\n",
            "epoch: 19568, dloss = 11.3, gloss = -3580.9, total_loss = -352.5 (0.68)\n",
            "epoch: 19569, dloss = 16.1, gloss = -3573.5, total_loss = -349.3 (0.67)\n",
            "epoch: 19570, dloss = 11.5, gloss = -3587.0, total_loss = -352.9 (0.88)\n",
            "epoch: 19571, dloss = 10.0, gloss = -3579.6, total_loss = -353.0 (0.86)\n",
            "epoch: 19572, dloss = 10.8, gloss = -3575.3, total_loss = -352.1 (0.67)\n",
            "epoch: 19573, dloss = 8.2, gloss = -3595.6, total_loss = -355.4 (0.67)\n",
            "epoch: 19574, dloss = 12.4, gloss = -3595.5, total_loss = -353.4 (0.68)\n",
            "epoch: 19575, dloss = 10.7, gloss = -3615.2, total_loss = -356.2 (0.69)\n",
            "epoch: 19576, dloss = 10.0, gloss = -3595.3, total_loss = -354.5 (0.65)\n",
            "epoch: 19577, dloss = 14.4, gloss = -3580.8, total_loss = -350.9 (0.68)\n",
            "epoch: 19578, dloss = 10.3, gloss = -3597.6, total_loss = -354.6 (0.68)\n",
            "epoch: 19579, dloss = 10.6, gloss = -3592.5, total_loss = -353.9 (0.65)\n",
            "epoch: 19580, dloss = 10.9, gloss = -3590.5, total_loss = -353.6 (0.70)\n",
            "epoch: 19581, dloss = 18.0, gloss = -3594.9, total_loss = -350.5 (0.68)\n",
            "epoch: 19582, dloss = 11.1, gloss = -3591.0, total_loss = -353.6 (0.65)\n",
            "epoch: 19583, dloss = 10.9, gloss = -3589.1, total_loss = -353.5 (0.71)\n",
            "epoch: 19584, dloss = 12.4, gloss = -3591.6, total_loss = -353.0 (0.69)\n",
            "epoch: 19585, dloss = 24.5, gloss = -3599.6, total_loss = -347.7 (0.68)\n",
            "epoch: 19586, dloss = 18.5, gloss = -3624.6, total_loss = -353.2 (0.96)\n",
            "epoch: 19587, dloss = 11.8, gloss = -3607.9, total_loss = -354.9 (0.77)\n",
            "epoch: 19588, dloss = 17.9, gloss = -3602.8, total_loss = -351.3 (0.67)\n",
            "epoch: 19589, dloss = 29.4, gloss = -3616.7, total_loss = -347.0 (0.67)\n",
            "epoch: 19590, dloss = 18.8, gloss = -3634.6, total_loss = -354.0 (0.69)\n",
            "epoch: 19591, dloss = 19.4, gloss = -3625.9, total_loss = -352.9 (0.68)\n",
            "epoch: 19592, dloss = 17.1, gloss = -3630.9, total_loss = -354.5 (0.67)\n",
            "epoch: 19593, dloss = 15.5, gloss = -3639.4, total_loss = -356.2 (0.68)\n",
            "epoch: 19594, dloss = 24.3, gloss = -3641.2, total_loss = -352.0 (0.68)\n",
            "epoch: 19595, dloss = 22.6, gloss = -3631.1, total_loss = -351.8 (0.68)\n",
            "epoch: 19596, dloss = 16.9, gloss = -3632.6, total_loss = -354.8 (0.67)\n",
            "epoch: 19597, dloss = 16.1, gloss = -3642.9, total_loss = -356.2 (0.68)\n",
            "epoch: 19598, dloss = 36.9, gloss = -3643.1, total_loss = -345.9 (0.68)\n",
            "epoch: 19599, dloss = 18.1, gloss = -3616.1, total_loss = -352.6 (0.69)\n",
            "epoch: 19600, dloss = 12.3, gloss = -3630.9, total_loss = -356.9 (0.69)\n",
            "epoch: 19601, dloss = 29.9, gloss = -3635.6, total_loss = -348.6 (0.77)\n",
            "epoch: 19602, dloss = 13.4, gloss = -3635.8, total_loss = -356.9 (0.99)\n",
            "epoch: 19603, dloss = 14.0, gloss = -3639.4, total_loss = -356.9 (0.68)\n",
            "epoch: 19604, dloss = 11.9, gloss = -3651.1, total_loss = -359.2 (0.68)\n",
            "epoch: 19605, dloss = 13.7, gloss = -3634.2, total_loss = -356.6 (0.69)\n",
            "epoch: 19606, dloss = 12.1, gloss = -3639.1, total_loss = -357.9 (0.67)\n",
            "epoch: 19607, dloss = 11.7, gloss = -3633.5, total_loss = -357.5 (0.68)\n",
            "epoch: 19608, dloss = 12.8, gloss = -3633.8, total_loss = -357.0 (0.67)\n",
            "epoch: 19609, dloss = 12.0, gloss = -3652.5, total_loss = -359.3 (0.68)\n",
            "epoch: 19610, dloss = 14.6, gloss = -3650.7, total_loss = -357.8 (0.68)\n",
            "epoch: 19611, dloss = 15.3, gloss = -3663.4, total_loss = -358.7 (0.69)\n",
            "epoch: 19612, dloss = 15.4, gloss = -3670.2, total_loss = -359.3 (0.71)\n",
            "epoch: 19613, dloss = 14.6, gloss = -3659.5, total_loss = -358.6 (0.66)\n",
            "epoch: 19614, dloss = 16.0, gloss = -3661.7, total_loss = -358.2 (0.66)\n",
            "epoch: 19615, dloss = 22.1, gloss = -3669.1, total_loss = -355.9 (0.71)\n",
            "epoch: 19616, dloss = 15.2, gloss = -3677.1, total_loss = -360.1 (0.90)\n",
            "epoch: 19617, dloss = 15.2, gloss = -3674.7, total_loss = -359.9 (1.13)\n",
            "epoch: 19618, dloss = 13.3, gloss = -3665.7, total_loss = -359.9 (0.76)\n",
            "epoch: 19619, dloss = 14.0, gloss = -3660.9, total_loss = -359.1 (0.72)\n",
            "epoch: 19620, dloss = 14.8, gloss = -3676.1, total_loss = -360.2 (0.72)\n",
            "epoch: 19621, dloss = 15.2, gloss = -3662.3, total_loss = -358.6 (0.70)\n",
            "epoch: 19622, dloss = 15.1, gloss = -3668.0, total_loss = -359.3 (0.69)\n",
            "epoch: 19623, dloss = 14.0, gloss = -3673.1, total_loss = -360.3 (0.69)\n",
            "epoch: 19624, dloss = 16.0, gloss = -3679.7, total_loss = -360.0 (0.67)\n",
            "epoch: 19625, dloss = 16.6, gloss = -3661.2, total_loss = -357.8 (0.68)\n",
            "epoch: 19626, dloss = 22.0, gloss = -3670.0, total_loss = -356.0 (0.98)\n",
            "epoch: 19627, dloss = 13.1, gloss = -3680.6, total_loss = -361.5 (0.79)\n",
            "epoch: 19628, dloss = 13.1, gloss = -3675.2, total_loss = -361.0 (0.70)\n",
            "epoch: 19629, dloss = 12.4, gloss = -3661.6, total_loss = -359.9 (0.68)\n",
            "epoch: 19630, dloss = 15.0, gloss = -3676.7, total_loss = -360.2 (0.74)\n",
            "epoch: 19631, dloss = 14.2, gloss = -3680.8, total_loss = -361.0 (1.08)\n",
            "epoch: 19632, dloss = 13.1, gloss = -3670.7, total_loss = -360.5 (1.09)\n",
            "epoch: 19633, dloss = 13.5, gloss = -3692.0, total_loss = -362.4 (0.69)\n",
            "epoch: 19634, dloss = 13.8, gloss = -3685.9, total_loss = -361.7 (0.69)\n",
            "epoch: 19635, dloss = 12.6, gloss = -3688.5, total_loss = -362.5 (0.67)\n",
            "epoch: 19636, dloss = 12.5, gloss = -3684.4, total_loss = -362.2 (0.69)\n",
            "epoch: 19637, dloss = 12.6, gloss = -3684.2, total_loss = -362.1 (0.71)\n",
            "epoch: 19638, dloss = 11.9, gloss = -3681.0, total_loss = -362.1 (0.69)\n",
            "epoch: 19639, dloss = 12.5, gloss = -3684.9, total_loss = -362.3 (0.70)\n",
            "epoch: 19640, dloss = 11.7, gloss = -3689.9, total_loss = -363.2 (0.71)\n",
            "epoch: 19641, dloss = 12.2, gloss = -3689.0, total_loss = -362.8 (0.68)\n",
            "epoch: 19642, dloss = 11.3, gloss = -3696.9, total_loss = -364.0 (0.68)\n",
            "epoch: 19643, dloss = 17.2, gloss = -3687.0, total_loss = -360.1 (0.68)\n",
            "epoch: 19644, dloss = 31.2, gloss = -3685.5, total_loss = -353.0 (0.69)\n",
            "epoch: 19645, dloss = 12.5, gloss = -3693.1, total_loss = -363.1 (0.68)\n",
            "epoch: 19646, dloss = 20.7, gloss = -3675.0, total_loss = -357.1 (0.67)\n",
            "epoch: 19647, dloss = 21.5, gloss = -3677.7, total_loss = -357.0 (0.88)\n",
            "epoch: 19648, dloss = 11.6, gloss = -3682.6, total_loss = -362.5 (0.85)\n",
            "epoch: 19649, dloss = 11.2, gloss = -3695.3, total_loss = -364.0 (0.69)\n",
            "epoch: 19650, dloss = 10.2, gloss = -3682.3, total_loss = -363.1 (0.67)\n",
            "epoch: 19651, dloss = 11.1, gloss = -3686.2, total_loss = -363.0 (0.68)\n",
            "epoch: 19652, dloss = 11.3, gloss = -3685.9, total_loss = -362.9 (0.69)\n",
            "epoch: 19653, dloss = 14.3, gloss = -3680.9, total_loss = -361.0 (0.70)\n",
            "epoch: 19654, dloss = 11.3, gloss = -3691.4, total_loss = -363.5 (0.71)\n",
            "epoch: 19655, dloss = 16.3, gloss = -3694.2, total_loss = -361.3 (0.68)\n",
            "epoch: 19656, dloss = 10.9, gloss = -3692.4, total_loss = -363.8 (0.70)\n",
            "epoch: 19657, dloss = 11.8, gloss = -3703.0, total_loss = -364.4 (0.70)\n",
            "epoch: 19658, dloss = 10.8, gloss = -3683.7, total_loss = -363.0 (0.67)\n",
            "epoch: 19659, dloss = 10.1, gloss = -3689.3, total_loss = -363.9 (0.67)\n",
            "epoch: 19660, dloss = 12.4, gloss = -3690.4, total_loss = -362.8 (0.69)\n",
            "epoch: 19661, dloss = 12.9, gloss = -3682.7, total_loss = -361.8 (0.67)\n",
            "epoch: 19662, dloss = 13.8, gloss = -3692.7, total_loss = -362.4 (0.69)\n",
            "epoch: 19663, dloss = 11.2, gloss = -3688.3, total_loss = -363.2 (1.04)\n",
            "epoch: 19664, dloss = 32.3, gloss = -3691.4, total_loss = -353.0 (0.69)\n",
            "epoch: 19665, dloss = 12.9, gloss = -3690.0, total_loss = -362.5 (0.67)\n",
            "epoch: 19666, dloss = 11.0, gloss = -3691.1, total_loss = -363.6 (0.68)\n",
            "epoch: 19667, dloss = 11.9, gloss = -3686.8, total_loss = -362.7 (0.69)\n",
            "epoch: 19668, dloss = 10.6, gloss = -3689.3, total_loss = -363.6 (0.66)\n",
            "epoch: 19669, dloss = 13.2, gloss = -3694.0, total_loss = -362.8 (0.69)\n",
            "epoch: 19670, dloss = 9.9, gloss = -3692.3, total_loss = -364.3 (0.68)\n",
            "epoch: 19671, dloss = 10.4, gloss = -3693.7, total_loss = -364.2 (0.66)\n",
            "epoch: 19672, dloss = 10.2, gloss = -3687.5, total_loss = -363.6 (0.71)\n",
            "epoch: 19673, dloss = 11.4, gloss = -3683.8, total_loss = -362.7 (0.70)\n",
            "epoch: 19674, dloss = 11.1, gloss = -3689.4, total_loss = -363.4 (0.71)\n",
            "epoch: 19675, dloss = 10.3, gloss = -3694.0, total_loss = -364.3 (0.70)\n",
            "epoch: 19676, dloss = 10.0, gloss = -3696.9, total_loss = -364.7 (0.72)\n",
            "epoch: 19677, dloss = 9.5, gloss = -3684.6, total_loss = -363.7 (0.68)\n",
            "epoch: 19678, dloss = 9.9, gloss = -3694.1, total_loss = -364.4 (0.85)\n",
            "epoch: 19679, dloss = 12.7, gloss = -3678.8, total_loss = -361.5 (0.88)\n",
            "epoch: 19680, dloss = 9.5, gloss = -3693.4, total_loss = -364.6 (0.70)\n",
            "epoch: 19681, dloss = 13.2, gloss = -3682.9, total_loss = -361.7 (0.69)\n",
            "epoch: 19682, dloss = 11.5, gloss = -3687.1, total_loss = -362.9 (0.70)\n",
            "epoch: 19683, dloss = 10.3, gloss = -3688.9, total_loss = -363.7 (0.71)\n",
            "epoch: 19684, dloss = 10.7, gloss = -3691.6, total_loss = -363.8 (0.71)\n",
            "epoch: 19685, dloss = 10.2, gloss = -3685.1, total_loss = -363.4 (0.68)\n",
            "epoch: 19686, dloss = 10.9, gloss = -3689.8, total_loss = -363.5 (0.72)\n",
            "epoch: 19687, dloss = 10.1, gloss = -3684.5, total_loss = -363.4 (0.68)\n",
            "epoch: 19688, dloss = 9.7, gloss = -3684.3, total_loss = -363.6 (0.66)\n",
            "epoch: 19689, dloss = 12.7, gloss = -3691.6, total_loss = -362.8 (0.71)\n",
            "epoch: 19690, dloss = 10.8, gloss = -3693.5, total_loss = -364.0 (0.70)\n",
            "epoch: 19691, dloss = 19.1, gloss = -3690.0, total_loss = -359.5 (0.70)\n",
            "epoch: 19692, dloss = 10.1, gloss = -3690.6, total_loss = -364.0 (0.70)\n",
            "epoch: 19693, dloss = 14.9, gloss = -3707.6, total_loss = -363.3 (0.75)\n",
            "epoch: 19694, dloss = 13.2, gloss = -3709.8, total_loss = -364.4 (1.02)\n",
            "epoch: 19695, dloss = 15.1, gloss = -3713.1, total_loss = -363.8 (0.68)\n",
            "epoch: 19696, dloss = 39.6, gloss = -3714.1, total_loss = -351.6 (0.70)\n",
            "epoch: 19697, dloss = 16.3, gloss = -3716.4, total_loss = -363.5 (0.70)\n",
            "epoch: 19698, dloss = 39.5, gloss = -3716.0, total_loss = -351.8 (0.69)\n",
            "epoch: 19699, dloss = 17.4, gloss = -3713.6, total_loss = -362.7 (0.71)\n",
            "epoch: 19700, dloss = 29.3, gloss = -3703.4, total_loss = -355.7 (0.71)\n",
            "epoch: 19701, dloss = 25.7, gloss = -3696.5, total_loss = -356.8 (0.67)\n",
            "epoch: 19702, dloss = 13.1, gloss = -3710.1, total_loss = -364.4 (0.68)\n",
            "epoch: 19703, dloss = 11.3, gloss = -3685.7, total_loss = -362.9 (0.69)\n",
            "epoch: 19704, dloss = 11.3, gloss = -3683.8, total_loss = -362.7 (0.67)\n",
            "epoch: 19705, dloss = 31.7, gloss = -3687.3, total_loss = -352.9 (0.70)\n",
            "epoch: 19706, dloss = 12.7, gloss = -3674.4, total_loss = -361.1 (0.69)\n",
            "epoch: 19707, dloss = 46.7, gloss = -3681.3, total_loss = -344.8 (0.71)\n",
            "epoch: 19708, dloss = 48.4, gloss = -3674.2, total_loss = -343.2 (0.68)\n",
            "epoch: 19709, dloss = 16.6, gloss = -3671.1, total_loss = -358.8 (0.95)\n",
            "epoch: 19710, dloss = 14.3, gloss = -3664.9, total_loss = -359.3 (0.79)\n",
            "epoch: 19711, dloss = 15.5, gloss = -3685.7, total_loss = -360.8 (0.68)\n",
            "epoch: 19712, dloss = 12.2, gloss = -3682.3, total_loss = -362.1 (0.68)\n",
            "epoch: 19713, dloss = 13.0, gloss = -3668.3, total_loss = -360.3 (0.68)\n",
            "epoch: 19714, dloss = 9.7, gloss = -3666.6, total_loss = -361.8 (0.67)\n",
            "epoch: 19715, dloss = 10.3, gloss = -3682.7, total_loss = -363.1 (0.66)\n",
            "epoch: 19716, dloss = 11.2, gloss = -3674.4, total_loss = -361.8 (0.68)\n",
            "epoch: 19717, dloss = 11.7, gloss = -3658.1, total_loss = -360.0 (0.68)\n",
            "epoch: 19718, dloss = 10.2, gloss = -3674.6, total_loss = -362.4 (0.69)\n",
            "epoch: 19719, dloss = 14.3, gloss = -3662.1, total_loss = -359.0 (0.69)\n",
            "epoch: 19720, dloss = 11.5, gloss = -3666.8, total_loss = -360.9 (0.69)\n",
            "epoch: 19721, dloss = 16.0, gloss = -3666.9, total_loss = -358.7 (0.67)\n",
            "epoch: 19722, dloss = 16.1, gloss = -3662.3, total_loss = -358.2 (0.68)\n",
            "epoch: 19723, dloss = 11.9, gloss = -3672.0, total_loss = -361.2 (0.69)\n",
            "epoch: 19724, dloss = 11.9, gloss = -3681.1, total_loss = -362.2 (0.71)\n",
            "epoch: 19725, dloss = 12.4, gloss = -3671.4, total_loss = -360.9 (1.02)\n",
            "epoch: 19726, dloss = 11.1, gloss = -3666.2, total_loss = -361.1 (0.70)\n",
            "epoch: 19727, dloss = 11.1, gloss = -3674.7, total_loss = -361.9 (0.70)\n",
            "epoch: 19728, dloss = 32.5, gloss = -3679.9, total_loss = -351.8 (0.68)\n",
            "epoch: 19729, dloss = 27.6, gloss = -3664.7, total_loss = -352.7 (0.66)\n",
            "epoch: 19730, dloss = 10.9, gloss = -3663.2, total_loss = -360.9 (0.67)\n",
            "epoch: 19731, dloss = 11.2, gloss = -3682.3, total_loss = -362.6 (0.66)\n",
            "epoch: 19732, dloss = 17.8, gloss = -3682.9, total_loss = -359.4 (0.69)\n",
            "epoch: 19733, dloss = 13.4, gloss = -3681.6, total_loss = -361.5 (0.71)\n",
            "epoch: 19734, dloss = 9.7, gloss = -3690.1, total_loss = -364.2 (0.66)\n",
            "epoch: 19735, dloss = 12.3, gloss = -3678.0, total_loss = -361.6 (0.70)\n",
            "epoch: 19736, dloss = 12.8, gloss = -3686.3, total_loss = -362.2 (0.69)\n",
            "epoch: 19737, dloss = 10.4, gloss = -3683.9, total_loss = -363.2 (0.68)\n",
            "epoch: 19738, dloss = 11.0, gloss = -3669.7, total_loss = -361.5 (0.69)\n",
            "epoch: 19739, dloss = 16.1, gloss = -3672.7, total_loss = -359.2 (0.68)\n",
            "epoch: 19740, dloss = 10.9, gloss = -3681.5, total_loss = -362.7 (0.83)\n",
            "epoch: 19741, dloss = 10.1, gloss = -3685.7, total_loss = -363.5 (0.88)\n",
            "epoch: 19742, dloss = 14.5, gloss = -3685.0, total_loss = -361.3 (0.67)\n",
            "epoch: 19743, dloss = 14.3, gloss = -3695.0, total_loss = -362.4 (0.67)\n",
            "epoch: 19744, dloss = 20.7, gloss = -3694.5, total_loss = -359.1 (0.66)\n",
            "epoch: 19745, dloss = 11.1, gloss = -3695.2, total_loss = -364.0 (0.71)\n",
            "epoch: 19746, dloss = 13.2, gloss = -3699.4, total_loss = -363.3 (0.68)\n",
            "epoch: 19747, dloss = 14.3, gloss = -3682.6, total_loss = -361.1 (0.67)\n",
            "epoch: 19748, dloss = 11.2, gloss = -3683.7, total_loss = -362.8 (0.69)\n",
            "epoch: 19749, dloss = 10.4, gloss = -3702.2, total_loss = -365.0 (0.68)\n",
            "epoch: 19750, dloss = 12.3, gloss = -3708.6, total_loss = -364.7 (0.68)\n",
            "epoch: 19751, dloss = 9.2, gloss = -3699.3, total_loss = -365.3 (0.68)\n",
            "epoch: 19752, dloss = 14.8, gloss = -3693.5, total_loss = -361.9 (0.67)\n",
            "epoch: 19753, dloss = 11.3, gloss = -3695.9, total_loss = -363.9 (0.68)\n",
            "epoch: 19754, dloss = 12.6, gloss = -3686.5, total_loss = -362.3 (0.69)\n",
            "epoch: 19755, dloss = 15.1, gloss = -3685.4, total_loss = -361.0 (0.69)\n",
            "epoch: 19756, dloss = 11.4, gloss = -3689.0, total_loss = -363.2 (0.96)\n",
            "epoch: 19757, dloss = 14.7, gloss = -3695.7, total_loss = -362.2 (0.79)\n",
            "epoch: 19758, dloss = 10.6, gloss = -3698.4, total_loss = -364.5 (0.69)\n",
            "epoch: 19759, dloss = 14.8, gloss = -3686.8, total_loss = -361.3 (0.69)\n",
            "epoch: 19760, dloss = 12.0, gloss = -3688.9, total_loss = -362.9 (0.67)\n",
            "epoch: 19761, dloss = 10.8, gloss = -3693.1, total_loss = -363.9 (0.69)\n",
            "epoch: 19762, dloss = 12.9, gloss = -3692.6, total_loss = -362.8 (0.71)\n",
            "epoch: 19763, dloss = 11.1, gloss = -3686.1, total_loss = -363.0 (0.69)\n",
            "epoch: 19764, dloss = 10.4, gloss = -3688.6, total_loss = -363.6 (0.69)\n",
            "epoch: 19765, dloss = 10.9, gloss = -3683.6, total_loss = -362.9 (0.69)\n",
            "epoch: 19766, dloss = 11.5, gloss = -3685.3, total_loss = -362.8 (0.67)\n",
            "epoch: 19767, dloss = 11.0, gloss = -3691.0, total_loss = -363.6 (0.69)\n",
            "epoch: 19768, dloss = 11.6, gloss = -3693.3, total_loss = -363.5 (0.70)\n",
            "epoch: 19769, dloss = 10.8, gloss = -3704.6, total_loss = -365.1 (0.69)\n",
            "epoch: 19770, dloss = 10.7, gloss = -3709.6, total_loss = -365.6 (0.69)\n",
            "epoch: 19771, dloss = 13.1, gloss = -3696.4, total_loss = -363.1 (0.80)\n",
            "epoch: 19772, dloss = 10.2, gloss = -3716.3, total_loss = -366.5 (0.92)\n",
            "epoch: 19773, dloss = 11.0, gloss = -3699.7, total_loss = -364.4 (0.66)\n",
            "epoch: 19774, dloss = 10.5, gloss = -3709.6, total_loss = -365.7 (0.69)\n",
            "epoch: 19775, dloss = 10.2, gloss = -3709.6, total_loss = -365.9 (0.68)\n",
            "epoch: 19776, dloss = 10.4, gloss = -3688.3, total_loss = -363.6 (0.68)\n",
            "epoch: 19777, dloss = 10.9, gloss = -3705.8, total_loss = -365.1 (0.69)\n",
            "epoch: 19778, dloss = 9.4, gloss = -3705.0, total_loss = -365.8 (0.70)\n",
            "epoch: 19779, dloss = 10.1, gloss = -3693.8, total_loss = -364.3 (0.67)\n",
            "epoch: 19780, dloss = 9.0, gloss = -3712.9, total_loss = -366.8 (0.65)\n",
            "epoch: 19781, dloss = 10.2, gloss = -3712.5, total_loss = -366.2 (0.69)\n",
            "epoch: 19782, dloss = 9.9, gloss = -3698.6, total_loss = -364.9 (0.69)\n",
            "epoch: 19783, dloss = 9.1, gloss = -3700.6, total_loss = -365.5 (0.65)\n",
            "epoch: 19784, dloss = 7.7, gloss = -3706.2, total_loss = -366.8 (0.69)\n",
            "epoch: 19785, dloss = 11.0, gloss = -3696.1, total_loss = -364.1 (0.69)\n",
            "epoch: 19786, dloss = 36.8, gloss = -3690.1, total_loss = -350.6 (0.68)\n",
            "epoch: 19787, dloss = 9.8, gloss = -3703.0, total_loss = -365.4 (0.90)\n",
            "epoch: 19788, dloss = 8.9, gloss = -3694.2, total_loss = -365.0 (0.88)\n",
            "epoch: 19789, dloss = 13.8, gloss = -3708.8, total_loss = -364.0 (0.68)\n",
            "epoch: 19790, dloss = 8.3, gloss = -3701.3, total_loss = -366.0 (0.68)\n",
            "epoch: 19791, dloss = 16.9, gloss = -3684.9, total_loss = -360.1 (0.68)\n",
            "epoch: 19792, dloss = 9.8, gloss = -3695.1, total_loss = -364.6 (0.68)\n",
            "epoch: 19793, dloss = 10.3, gloss = -3696.4, total_loss = -364.5 (0.69)\n",
            "epoch: 19794, dloss = 11.2, gloss = -3702.2, total_loss = -364.6 (0.68)\n",
            "epoch: 19795, dloss = 14.1, gloss = -3712.0, total_loss = -364.2 (0.68)\n",
            "epoch: 19796, dloss = 10.4, gloss = -3704.5, total_loss = -365.2 (0.66)\n",
            "epoch: 19797, dloss = 11.1, gloss = -3694.7, total_loss = -363.9 (0.68)\n",
            "epoch: 19798, dloss = 12.6, gloss = -3685.9, total_loss = -362.3 (0.66)\n",
            "epoch: 19799, dloss = 10.4, gloss = -3690.7, total_loss = -363.9 (0.77)\n",
            "epoch: 19800, dloss = 12.4, gloss = -3705.9, total_loss = -364.4 (1.26)\n",
            "epoch: 19801, dloss = 9.9, gloss = -3710.6, total_loss = -366.1 (1.29)\n",
            "epoch: 19802, dloss = 9.3, gloss = -3715.9, total_loss = -367.0 (1.05)\n",
            "epoch: 19803, dloss = 9.8, gloss = -3720.3, total_loss = -367.1 (0.69)\n",
            "epoch: 19804, dloss = 9.9, gloss = -3716.3, total_loss = -366.7 (0.70)\n",
            "epoch: 19805, dloss = 11.1, gloss = -3719.3, total_loss = -366.4 (0.69)\n",
            "epoch: 19806, dloss = 20.2, gloss = -3726.4, total_loss = -362.5 (0.70)\n",
            "epoch: 19807, dloss = 11.1, gloss = -3724.0, total_loss = -366.9 (0.67)\n",
            "epoch: 19808, dloss = 12.0, gloss = -3729.4, total_loss = -367.0 (0.71)\n",
            "epoch: 19809, dloss = 200.8, gloss = -3733.1, total_loss = -272.9 (0.68)\n",
            "epoch: 19810, dloss = 13.9, gloss = -3713.7, total_loss = -364.4 (0.69)\n",
            "epoch: 19811, dloss = 10.3, gloss = -3725.8, total_loss = -367.4 (0.72)\n",
            "epoch: 19812, dloss = 11.5, gloss = -3733.4, total_loss = -367.6 (0.69)\n",
            "epoch: 19813, dloss = 14.0, gloss = -3734.0, total_loss = -366.4 (0.68)\n",
            "epoch: 19814, dloss = 8.4, gloss = -3724.6, total_loss = -368.2 (0.68)\n",
            "epoch: 19815, dloss = 10.1, gloss = -3729.0, total_loss = -367.8 (0.68)\n",
            "epoch: 19816, dloss = 14.1, gloss = -3735.0, total_loss = -366.5 (0.67)\n",
            "epoch: 19817, dloss = 9.4, gloss = -3733.9, total_loss = -368.7 (0.98)\n",
            "epoch: 19818, dloss = 10.3, gloss = -3729.3, total_loss = -367.8 (0.75)\n",
            "epoch: 19819, dloss = 9.9, gloss = -3736.6, total_loss = -368.7 (0.68)\n",
            "epoch: 19820, dloss = 10.5, gloss = -3724.9, total_loss = -367.3 (0.68)\n",
            "epoch: 19821, dloss = 10.1, gloss = -3740.2, total_loss = -369.0 (0.68)\n",
            "epoch: 19822, dloss = 10.0, gloss = -3738.0, total_loss = -368.8 (0.69)\n",
            "epoch: 19823, dloss = 11.6, gloss = -3727.9, total_loss = -367.0 (0.68)\n",
            "epoch: 19824, dloss = 10.3, gloss = -3738.2, total_loss = -368.7 (0.69)\n",
            "epoch: 19825, dloss = 10.7, gloss = -3732.5, total_loss = -367.9 (0.68)\n",
            "epoch: 19826, dloss = 9.9, gloss = -3725.0, total_loss = -367.6 (0.68)\n",
            "epoch: 19827, dloss = 13.1, gloss = -3737.6, total_loss = -367.2 (0.68)\n",
            "epoch: 19828, dloss = 11.3, gloss = -3719.9, total_loss = -366.3 (0.69)\n",
            "epoch: 19829, dloss = 10.0, gloss = -3730.4, total_loss = -368.0 (0.67)\n",
            "epoch: 19830, dloss = 77.8, gloss = -3718.9, total_loss = -333.0 (0.67)\n",
            "epoch: 19831, dloss = 10.4, gloss = -3719.5, total_loss = -366.7 (0.70)\n",
            "epoch: 19832, dloss = 9.3, gloss = -3726.8, total_loss = -368.0 (0.78)\n",
            "epoch: 19833, dloss = 9.7, gloss = -3738.3, total_loss = -369.0 (0.95)\n",
            "epoch: 19834, dloss = 9.0, gloss = -3735.5, total_loss = -369.1 (0.70)\n",
            "epoch: 19835, dloss = 9.7, gloss = -3736.2, total_loss = -368.8 (0.68)\n",
            "epoch: 19836, dloss = 8.7, gloss = -3734.3, total_loss = -369.1 (0.69)\n",
            "epoch: 19837, dloss = 23.3, gloss = -3729.6, total_loss = -361.3 (0.67)\n",
            "epoch: 19838, dloss = 10.4, gloss = -3731.7, total_loss = -368.0 (0.68)\n",
            "epoch: 19839, dloss = 8.9, gloss = -3741.0, total_loss = -369.7 (0.68)\n",
            "epoch: 19840, dloss = 9.4, gloss = -3733.0, total_loss = -368.6 (0.67)\n",
            "epoch: 19841, dloss = 11.3, gloss = -3728.2, total_loss = -367.2 (0.68)\n",
            "epoch: 19842, dloss = 11.4, gloss = -3723.4, total_loss = -366.7 (1.26)\n",
            "epoch: 19843, dloss = 9.3, gloss = -3728.8, total_loss = -368.2 (0.68)\n",
            "epoch: 19844, dloss = 18.1, gloss = -3726.9, total_loss = -363.7 (0.69)\n",
            "epoch: 19845, dloss = 9.3, gloss = -3722.8, total_loss = -367.6 (0.72)\n",
            "epoch: 19846, dloss = 8.8, gloss = -3730.9, total_loss = -368.7 (0.71)\n",
            "epoch: 19847, dloss = 12.0, gloss = -3730.4, total_loss = -367.0 (0.89)\n",
            "epoch: 19848, dloss = 10.8, gloss = -3718.1, total_loss = -366.4 (0.88)\n",
            "epoch: 19849, dloss = 9.3, gloss = -3721.6, total_loss = -367.5 (0.70)\n",
            "epoch: 19850, dloss = 8.5, gloss = -3734.5, total_loss = -369.2 (0.74)\n",
            "epoch: 19851, dloss = 9.9, gloss = -3727.6, total_loss = -367.8 (0.69)\n",
            "epoch: 19852, dloss = 9.1, gloss = -3737.7, total_loss = -369.2 (0.69)\n",
            "epoch: 19853, dloss = 10.0, gloss = -3730.6, total_loss = -368.1 (0.69)\n",
            "epoch: 19854, dloss = 23.6, gloss = -3737.2, total_loss = -361.9 (0.71)\n",
            "epoch: 19855, dloss = 9.5, gloss = -3730.6, total_loss = -368.3 (0.69)\n",
            "epoch: 19856, dloss = 9.5, gloss = -3730.6, total_loss = -368.3 (0.66)\n",
            "epoch: 19857, dloss = 9.2, gloss = -3721.6, total_loss = -367.6 (0.69)\n",
            "epoch: 19858, dloss = 8.4, gloss = -3737.7, total_loss = -369.6 (0.69)\n",
            "epoch: 19859, dloss = 9.1, gloss = -3726.2, total_loss = -368.1 (0.69)\n",
            "epoch: 19860, dloss = 9.1, gloss = -3734.7, total_loss = -368.9 (0.70)\n",
            "epoch: 19861, dloss = 10.6, gloss = -3732.2, total_loss = -367.9 (0.69)\n",
            "epoch: 19862, dloss = 8.4, gloss = -3735.2, total_loss = -369.3 (0.74)\n",
            "epoch: 19863, dloss = 8.6, gloss = -3723.4, total_loss = -368.1 (0.99)\n",
            "epoch: 19864, dloss = 9.3, gloss = -3722.6, total_loss = -367.6 (0.68)\n",
            "epoch: 19865, dloss = 9.6, gloss = -3725.8, total_loss = -367.8 (0.68)\n",
            "epoch: 19866, dloss = 10.0, gloss = -3723.8, total_loss = -367.4 (0.70)\n",
            "epoch: 19867, dloss = 9.4, gloss = -3724.8, total_loss = -367.8 (0.84)\n",
            "epoch: 19868, dloss = 9.1, gloss = -3726.9, total_loss = -368.2 (0.94)\n",
            "epoch: 19869, dloss = 9.3, gloss = -3727.4, total_loss = -368.1 (0.69)\n",
            "epoch: 19870, dloss = 9.2, gloss = -3730.5, total_loss = -368.4 (0.70)\n",
            "epoch: 19871, dloss = 10.3, gloss = -3729.0, total_loss = -367.7 (0.68)\n",
            "epoch: 19872, dloss = 7.8, gloss = -3727.8, total_loss = -368.9 (0.67)\n",
            "epoch: 19873, dloss = 9.2, gloss = -3735.8, total_loss = -369.0 (0.68)\n",
            "epoch: 19874, dloss = 8.7, gloss = -3741.6, total_loss = -369.8 (0.69)\n",
            "epoch: 19875, dloss = 10.2, gloss = -3732.1, total_loss = -368.1 (0.68)\n",
            "epoch: 19876, dloss = 9.5, gloss = -3726.7, total_loss = -367.9 (0.69)\n",
            "epoch: 19877, dloss = 9.3, gloss = -3725.0, total_loss = -367.9 (0.77)\n",
            "epoch: 19878, dloss = 10.1, gloss = -3729.8, total_loss = -367.9 (1.11)\n",
            "epoch: 19879, dloss = 9.7, gloss = -3734.5, total_loss = -368.6 (0.92)\n",
            "epoch: 19880, dloss = 10.0, gloss = -3725.4, total_loss = -367.5 (0.67)\n",
            "epoch: 19881, dloss = 8.9, gloss = -3720.5, total_loss = -367.6 (0.68)\n",
            "epoch: 19882, dloss = 8.6, gloss = -3736.8, total_loss = -369.4 (0.68)\n",
            "epoch: 19883, dloss = 8.3, gloss = -3737.7, total_loss = -369.6 (0.68)\n",
            "epoch: 19884, dloss = 8.7, gloss = -3731.9, total_loss = -368.8 (0.68)\n",
            "epoch: 19885, dloss = 8.6, gloss = -3726.3, total_loss = -368.3 (0.69)\n",
            "epoch: 19886, dloss = 8.4, gloss = -3730.6, total_loss = -368.9 (0.67)\n",
            "epoch: 19887, dloss = 9.0, gloss = -3723.2, total_loss = -367.8 (0.68)\n",
            "epoch: 19888, dloss = 10.4, gloss = -3728.0, total_loss = -367.6 (0.67)\n",
            "epoch: 19889, dloss = 9.3, gloss = -3728.4, total_loss = -368.2 (0.68)\n",
            "epoch: 19890, dloss = 9.6, gloss = -3726.2, total_loss = -367.8 (0.69)\n",
            "epoch: 19891, dloss = 8.5, gloss = -3720.1, total_loss = -367.8 (0.68)\n",
            "epoch: 19892, dloss = 9.4, gloss = -3722.2, total_loss = -367.5 (0.70)\n",
            "epoch: 19893, dloss = 8.8, gloss = -3712.3, total_loss = -366.8 (0.89)\n",
            "epoch: 19894, dloss = 9.5, gloss = -3723.3, total_loss = -367.6 (0.83)\n",
            "epoch: 19895, dloss = 9.2, gloss = -3723.4, total_loss = -367.7 (0.67)\n",
            "epoch: 19896, dloss = 10.2, gloss = -3710.3, total_loss = -365.9 (0.70)\n",
            "epoch: 19897, dloss = 7.4, gloss = -3718.8, total_loss = -368.2 (0.69)\n",
            "epoch: 19898, dloss = 10.2, gloss = -3722.9, total_loss = -367.2 (0.67)\n",
            "epoch: 19899, dloss = 9.2, gloss = -3729.0, total_loss = -368.3 (0.69)\n",
            "epoch: 19900, dloss = 8.5, gloss = -3728.0, total_loss = -368.6 (0.71)\n",
            "epoch: 19901, dloss = 9.1, gloss = -3726.5, total_loss = -368.1 (0.68)\n",
            "epoch: 19902, dloss = 15.3, gloss = -3738.7, total_loss = -366.2 (0.68)\n",
            "epoch: 19903, dloss = 20.2, gloss = -3736.7, total_loss = -363.6 (0.67)\n",
            "epoch: 19904, dloss = 9.0, gloss = -3740.4, total_loss = -369.5 (0.70)\n",
            "epoch: 19905, dloss = 10.5, gloss = -3734.2, total_loss = -368.2 (0.72)\n",
            "epoch: 19906, dloss = 13.8, gloss = -3748.5, total_loss = -368.0 (0.69)\n",
            "epoch: 19907, dloss = 16.9, gloss = -3745.7, total_loss = -366.1 (0.70)\n",
            "epoch: 19908, dloss = 9.1, gloss = -3736.7, total_loss = -369.1 (0.77)\n",
            "epoch: 19909, dloss = 8.9, gloss = -3738.7, total_loss = -369.4 (0.98)\n",
            "epoch: 19910, dloss = 7.9, gloss = -3730.3, total_loss = -369.1 (0.67)\n",
            "epoch: 19911, dloss = 9.6, gloss = -3727.9, total_loss = -368.0 (0.69)\n",
            "epoch: 19912, dloss = 8.3, gloss = -3732.0, total_loss = -369.1 (0.68)\n",
            "epoch: 19913, dloss = 9.4, gloss = -3731.0, total_loss = -368.4 (0.68)\n",
            "epoch: 19914, dloss = 8.9, gloss = -3737.7, total_loss = -369.3 (0.70)\n",
            "epoch: 19915, dloss = 9.5, gloss = -3729.4, total_loss = -368.2 (0.69)\n",
            "epoch: 19916, dloss = 8.4, gloss = -3730.5, total_loss = -368.8 (0.70)\n",
            "epoch: 19917, dloss = 21.0, gloss = -3736.8, total_loss = -363.2 (0.69)\n",
            "epoch: 19918, dloss = 8.4, gloss = -3737.1, total_loss = -369.5 (0.71)\n",
            "epoch: 19919, dloss = 8.8, gloss = -3738.6, total_loss = -369.4 (0.67)\n",
            "epoch: 19920, dloss = 8.3, gloss = -3730.7, total_loss = -368.9 (0.69)\n",
            "epoch: 19921, dloss = 49.0, gloss = -3736.5, total_loss = -349.2 (0.70)\n",
            "epoch: 19922, dloss = 8.4, gloss = -3736.5, total_loss = -369.4 (0.71)\n",
            "epoch: 19923, dloss = 9.1, gloss = -3730.7, total_loss = -368.5 (0.69)\n",
            "epoch: 19924, dloss = 8.9, gloss = -3727.3, total_loss = -368.3 (0.95)\n",
            "epoch: 19925, dloss = 9.7, gloss = -3729.5, total_loss = -368.1 (0.85)\n",
            "epoch: 19926, dloss = 9.4, gloss = -3735.9, total_loss = -368.9 (1.04)\n",
            "epoch: 19927, dloss = 7.3, gloss = -3725.3, total_loss = -368.9 (0.78)\n",
            "epoch: 19928, dloss = 91.7, gloss = -3723.4, total_loss = -326.5 (0.67)\n",
            "epoch: 19929, dloss = 8.4, gloss = -3724.4, total_loss = -368.3 (0.70)\n",
            "epoch: 19930, dloss = 8.7, gloss = -3716.2, total_loss = -367.3 (0.71)\n",
            "epoch: 19931, dloss = 9.0, gloss = -3730.8, total_loss = -368.6 (0.68)\n",
            "epoch: 19932, dloss = 8.6, gloss = -3743.2, total_loss = -370.0 (0.69)\n",
            "epoch: 19933, dloss = 8.9, gloss = -3734.8, total_loss = -369.0 (0.69)\n",
            "epoch: 19934, dloss = 8.8, gloss = -3722.1, total_loss = -367.8 (0.69)\n",
            "epoch: 19935, dloss = 10.3, gloss = -3736.2, total_loss = -368.5 (0.70)\n",
            "epoch: 19936, dloss = 18.2, gloss = -3735.0, total_loss = -364.4 (0.69)\n",
            "epoch: 19937, dloss = 9.4, gloss = -3743.2, total_loss = -369.6 (0.70)\n",
            "epoch: 19938, dloss = 66.0, gloss = -3725.8, total_loss = -339.6 (0.68)\n",
            "epoch: 19939, dloss = 10.0, gloss = -3734.5, total_loss = -368.5 (1.01)\n",
            "epoch: 19940, dloss = 10.1, gloss = -3739.1, total_loss = -368.8 (0.74)\n",
            "epoch: 19941, dloss = 8.7, gloss = -3745.3, total_loss = -370.2 (0.70)\n",
            "epoch: 19942, dloss = 9.2, gloss = -3745.7, total_loss = -370.0 (0.67)\n",
            "epoch: 19943, dloss = 98.3, gloss = -3731.0, total_loss = -323.9 (0.68)\n",
            "epoch: 19944, dloss = 9.3, gloss = -3742.8, total_loss = -369.6 (0.69)\n",
            "epoch: 19945, dloss = 8.2, gloss = -3736.7, total_loss = -369.6 (0.67)\n",
            "epoch: 19946, dloss = 7.6, gloss = -3729.9, total_loss = -369.2 (0.68)\n",
            "epoch: 19947, dloss = 7.8, gloss = -3725.3, total_loss = -368.6 (0.68)\n",
            "epoch: 19948, dloss = 7.7, gloss = -3726.7, total_loss = -368.8 (0.69)\n",
            "epoch: 19949, dloss = 9.2, gloss = -3743.7, total_loss = -369.8 (0.69)\n",
            "epoch: 19950, dloss = 8.4, gloss = -3746.7, total_loss = -370.5 (0.67)\n",
            "epoch: 19951, dloss = 9.2, gloss = -3745.1, total_loss = -369.9 (0.69)\n",
            "epoch: 19952, dloss = 8.6, gloss = -3743.4, total_loss = -370.0 (0.69)\n",
            "epoch: 19953, dloss = 8.4, gloss = -3737.6, total_loss = -369.5 (0.69)\n",
            "epoch: 19954, dloss = 18.5, gloss = -3745.5, total_loss = -365.3 (0.79)\n",
            "epoch: 19955, dloss = 9.5, gloss = -3748.4, total_loss = -370.1 (0.94)\n",
            "epoch: 19956, dloss = 9.0, gloss = -3746.0, total_loss = -370.1 (0.70)\n",
            "epoch: 19957, dloss = 9.6, gloss = -3754.0, total_loss = -370.6 (0.69)\n",
            "epoch: 19958, dloss = 8.9, gloss = -3753.2, total_loss = -370.9 (0.68)\n",
            "epoch: 19959, dloss = 8.4, gloss = -3759.8, total_loss = -371.8 (0.68)\n",
            "epoch: 19960, dloss = 8.6, gloss = -3757.7, total_loss = -371.5 (0.67)\n",
            "epoch: 19961, dloss = 8.4, gloss = -3750.5, total_loss = -370.9 (0.65)\n",
            "epoch: 19962, dloss = 8.2, gloss = -3750.9, total_loss = -371.0 (0.67)\n",
            "epoch: 19963, dloss = 8.2, gloss = -3747.2, total_loss = -370.6 (0.70)\n",
            "epoch: 19964, dloss = 8.5, gloss = -3752.2, total_loss = -370.9 (0.68)\n",
            "epoch: 19965, dloss = 9.3, gloss = -3755.3, total_loss = -370.9 (0.68)\n",
            "epoch: 19966, dloss = 28.3, gloss = -3760.6, total_loss = -361.9 (0.69)\n",
            "epoch: 19967, dloss = 11.1, gloss = -3748.5, total_loss = -369.3 (0.67)\n",
            "epoch: 19968, dloss = 105.7, gloss = -3753.2, total_loss = -322.5 (0.70)\n",
            "epoch: 19969, dloss = 14.3, gloss = -3751.8, total_loss = -368.0 (0.68)\n",
            "epoch: 19970, dloss = 18.3, gloss = -3749.5, total_loss = -365.8 (0.89)\n",
            "epoch: 19971, dloss = 9.1, gloss = -3750.5, total_loss = -370.5 (0.88)\n",
            "epoch: 19972, dloss = 7.9, gloss = -3755.2, total_loss = -371.6 (0.69)\n",
            "epoch: 19973, dloss = 7.3, gloss = -3759.9, total_loss = -372.3 (0.70)\n",
            "epoch: 19974, dloss = 9.1, gloss = -3755.5, total_loss = -371.0 (0.69)\n",
            "epoch: 19975, dloss = 15.7, gloss = -3754.6, total_loss = -367.6 (0.70)\n",
            "epoch: 19976, dloss = 8.0, gloss = -3755.4, total_loss = -371.5 (0.68)\n",
            "epoch: 19977, dloss = 15.0, gloss = -3738.2, total_loss = -366.3 (0.69)\n",
            "epoch: 19978, dloss = 25.6, gloss = -3753.4, total_loss = -362.5 (0.68)\n",
            "epoch: 19979, dloss = 9.7, gloss = -3749.8, total_loss = -370.1 (0.70)\n",
            "epoch: 19980, dloss = 11.8, gloss = -3744.6, total_loss = -368.5 (0.67)\n",
            "epoch: 19981, dloss = 9.0, gloss = -3751.5, total_loss = -370.6 (0.70)\n",
            "epoch: 19982, dloss = 8.9, gloss = -3748.7, total_loss = -370.4 (0.69)\n",
            "epoch: 19983, dloss = 12.6, gloss = -3743.4, total_loss = -368.1 (0.69)\n",
            "epoch: 19984, dloss = 9.1, gloss = -3738.6, total_loss = -369.3 (0.69)\n",
            "epoch: 19985, dloss = 8.5, gloss = -3743.3, total_loss = -370.1 (0.72)\n",
            "epoch: 19986, dloss = 9.1, gloss = -3743.1, total_loss = -369.8 (1.01)\n",
            "epoch: 19987, dloss = 8.9, gloss = -3741.6, total_loss = -369.7 (0.68)\n",
            "epoch: 19988, dloss = 9.5, gloss = -3754.1, total_loss = -370.7 (0.69)\n",
            "epoch: 19989, dloss = 8.4, gloss = -3756.0, total_loss = -371.4 (0.69)\n",
            "epoch: 19990, dloss = 8.2, gloss = -3745.9, total_loss = -370.5 (0.68)\n",
            "epoch: 19991, dloss = 9.9, gloss = -3731.6, total_loss = -368.2 (0.67)\n",
            "epoch: 19992, dloss = 8.2, gloss = -3749.4, total_loss = -370.9 (0.69)\n",
            "epoch: 19993, dloss = 7.6, gloss = -3743.1, total_loss = -370.5 (0.69)\n",
            "epoch: 19994, dloss = 8.2, gloss = -3740.3, total_loss = -369.9 (0.68)\n",
            "epoch: 19995, dloss = 9.3, gloss = -3737.1, total_loss = -369.0 (0.69)\n",
            "epoch: 19996, dloss = 7.7, gloss = -3732.6, total_loss = -369.4 (0.68)\n",
            "epoch: 19997, dloss = 10.2, gloss = -3729.5, total_loss = -367.9 (0.68)\n",
            "epoch: 19998, dloss = 8.9, gloss = -3734.1, total_loss = -369.0 (0.70)\n",
            "epoch: 19999, dloss = 8.0, gloss = -3743.5, total_loss = -370.3 (0.68)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Results\n",
        "\n",
        "### Generate a new set of data\n",
        "\n",
        "Remember to set the generator as not trainable"
      ],
      "metadata": {
        "id": "7-7TRZCW4pwh"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "checkpoint.restore(manager.latest_checkpoint).expect_partial().assert_consumed()\n",
        "\n",
        "decoder.trainable = False\n",
        "generator.trainable = False\n",
        "\n",
        "def g_step_generator():\n",
        "    z = tf.random.normal(shape=[train_data.shape[0] * 1000, Z_DIM])\n",
        "    synthetic = decoder(generator(z))\n",
        "\n",
        "    return synthetic\n",
        "\n",
        "synthetic = decoder(generator(z))\n",
        "syn = discriminator(synthetic)\n",
        "\n",
        "np.save(\"/content/AE-WGAN-GP/Generated_data/output_file_10000.npy\", synthetic)"
      ],
      "metadata": {
        "id": "sfv10-s0Avq3"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Preprocesing\n",
        "\n",
        "### Load generated and real data"
      ],
      "metadata": {
        "id": "6woOL_KD3gJ1"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "output_data = np.load(\"/content/AE-WGAN-GP/Generated_data/output_file_10000.npy\", allow_pickle=True)\n",
        "true_data = np.load(\"/content/AE-WGAN-GP/Input_data/D3_events_ALL_OUTCOMES_ML.npy\", allow_pickle=True)\n",
        "\n",
        "output_data = np.round(output_data)"
      ],
      "metadata": {
        "id": "phlnAKtwHbae"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='4'>Cleaning</font>\n",
        "\n",
        "Some observation have only values 0.\n",
        "\n",
        "We should remove them before using the dataset"
      ],
      "metadata": {
        "id": "s4vJ1ag97FtU"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "col_totals = pd.DataFrame(true_data.sum(0))\n",
        "\n",
        "col_totals_small = pd.DataFrame(output_data)\n",
        "\n",
        "proportion_0 = (col_totals_small[(col_totals_small.T == 0).all()].shape)[0] / (col_totals_small.shape)[0]\n",
        "format_str = 'Percentage of rows with only 0: %.2f%%'\n",
        "print(format_str % (proportion_0 * 100))\n",
        "\n",
        "col_totals_small = col_totals_small[(col_totals_small.T != 0).any()]"
      ],
      "metadata": {
        "id": "QPC6e5Dc7FcN",
        "outputId": "2d9c2b76-123e-47f7-8aae-117c7611975b",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Percentage of rows with only 0: 1.93%\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Compute statistics by column\n",
        "\n",
        "### Marginals\n",
        "\n",
        "The script reproduce well most of the variables.\n",
        "\n",
        "D_PANCRACUTE_AESI, C_MYOCARD_AESI and I_INFLUENZA_AESI are sligthly underepreentend in the generated data\n",
        "\n",
        "E_DM1_AESI on the other hand is over represented"
      ],
      "metadata": {
        "id": "gAz8194i35j1"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "size_factor = math.floor(col_totals_small.shape[0] / true_data.shape[0])\n",
        "\n",
        "col_totals_small = col_totals_small.head(size_factor * true_data.shape[0])\n",
        "col_totals_small = pd.DataFrame(col_totals_small.sum(0) / size_factor)\n",
        "\n",
        "pd_df = col_totals.join(col_totals_small, lsuffix='_real', rsuffix='_synthetic').fillna(0)\n",
        "pd_df['Varnames'] = [\"B_COAGDIS_AESI\", \"C_ARRH_AESI\", \"C_CAD_AESI\", \"C_MYOCARD_AESI\", \"C_VALVULAR_AESI\", \"DEATH\", \"D_PANCRACUTE_AESI\",\n",
        "     \"E_DM1_AESI\", \"E_GOUT_AESI\", \"G_KIACUTE_AESI\", \"G_UTI_AESI\", \"I_INFLUENZA_AESI\", \"M_FRACTURES_AESI\",\n",
        "     \"M_OSTEOARTHRITIS_AESI\", \"N_STROKEHEMO_AESI\", \"SO_OTITISEXT_AESI\", \"V_THROMBOSISARTERIALALGOR_AESI\",\n",
        "     \"V_VTEALGORITHM_AESI\"]\n",
        "pd_df = pd_df.set_index('Varnames')\n",
        "pd_df[\"diff_perc\"] = round((pd_df['0_real'] - pd_df['0_synthetic']) / pd_df['0_real'] * 100, 2)\n",
        "pp.pprint(pd_df)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aLs5Uh7v34_4",
        "outputId": "7861fa25-2905-4294-fb17-86344ccc0212"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "                                0_real  0_synthetic  diff_perc\n",
            "Varnames                                                      \n",
            "B_COAGDIS_AESI                    73.0    73.715309      -0.98\n",
            "C_ARRH_AESI                       65.0    69.045921      -6.22\n",
            "C_CAD_AESI                        31.0    31.479591      -1.55\n",
            "C_MYOCARD_AESI                    20.0    17.986734      10.07\n",
            "C_VALVULAR_AESI                   50.0    51.130611      -2.26\n",
            "DEATH                            258.0   253.684692       1.67\n",
            "D_PANCRACUTE_AESI                 13.0     9.452041      27.29\n",
            "E_DM1_AESI                        38.0    44.208164     -16.34\n",
            "E_GOUT_AESI                       27.0    25.344898       6.13\n",
            "G_KIACUTE_AESI                    15.0    14.710204       1.93\n",
            "G_UTI_AESI                        23.0    23.219387      -0.95\n",
            "I_INFLUENZA_AESI                  16.0    14.138776      11.63\n",
            "M_FRACTURES_AESI                  60.0    59.522449       0.80\n",
            "M_OSTEOARTHRITIS_AESI             28.0    27.116327       3.16\n",
            "N_STROKEHEMO_AESI                 10.0     9.992857       0.07\n",
            "SO_OTITISEXT_AESI                 36.0    37.114285      -3.10\n",
            "V_THROMBOSISARTERIALALGOR_AESI    45.0    45.730614      -1.62\n",
            "V_VTEALGORITHM_AESI               19.0    19.263266      -1.39\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='4'>Mean square error (MSE)</font>\n",
        "\n",
        "Very low"
      ],
      "metadata": {
        "id": "yn987c4F6FIY"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "format_str = 'MSE: %.2f'\n",
        "print(format_str % (round(np.square(pd_df['0_real'] - pd_df['0_synthetic']).mean(), 2)))"
      ],
      "metadata": {
        "id": "BQYE0ybW6F_z",
        "outputId": "1432f9a1-f2d7-46ee-a922-0722aca321c2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "MSE: 5.63\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Compute statistics by row\n",
        "\n",
        "### Marginals\n",
        "\n",
        "The script reproduce well the numgber of occurences of the variables"
      ],
      "metadata": {
        "id": "qNHHlhDu-WvS"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "row_totals = true_data.sum(1)\n",
        "row_totals_small = pd.DataFrame(output_data.sum(1))\n",
        "\n",
        "unique, counts = np.unique(row_totals, return_counts=True)\n",
        "row_totals_matrix = pd.DataFrame(counts, unique)\n",
        "unique, counts = np.unique(row_totals_small, return_counts=True)\n",
        "row_totals_small_matrix = pd.DataFrame(counts / size_factor, unique)\n",
        "\n",
        "pd_df = row_totals_matrix.join(row_totals_small_matrix, lsuffix='_real', rsuffix='_synthetic').fillna(0)\n",
        "pd_df[\"diff_perc\"] = round((pd_df['0_real'] - pd_df['0_synthetic']) / pd_df['0_real'] * 100, 2)\n",
        "pp.pprint(pd_df)"
      ],
      "metadata": {
        "id": "QqPs54L35uq9",
        "outputId": "0fe6f98d-d476-4d2e-f34b-29e1b37edfbc",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "     0_real  0_synthetic  diff_perc\n",
            "1.0     639   639.180612      -0.03\n",
            "2.0      52    52.792857      -1.52\n",
            "3.0      28    27.562245       1.56\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='4'>Mean square error (MSE)</font>\n",
        "\n",
        "Very low"
      ],
      "metadata": {
        "id": "Qumo6FcB5vFv"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "format_str = 'MSE: %.2f'\n",
        "print(format_str % (round(np.square(pd_df['0_real'] - pd_df['0_synthetic']).mean(), 2)))"
      ],
      "metadata": {
        "id": "XIRYNx6D-oxA",
        "outputId": "a780a950-1ffc-449c-ccf3-c0ad0107dfe2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "MSE: 0.28\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Loss history\n",
        "\n",
        "## Discriminator/Critic Loss"
      ],
      "metadata": {
        "id": "1RgghbR2_Q6a"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "with open(os.path.join(checkpoint_directory, \"dloss.pkl\"), 'rb') as f:\n",
        "  dloss_list = pickle.load(f)\n",
        "fig, ax = plt.subplots()\n",
        "ax.plot(dloss_list)\n",
        "plt.title('Discriminator/Critic Loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend('d-loss', loc='lower right')\n",
        "plt.show()\n",
        "fig, ax = plt.subplots()\n",
        "ax.plot(dloss_list)\n",
        "ax.set_ylim(-100, 100)\n",
        "plt.title('Discriminator/Critic Loss (Zoom in)')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend('d-loss', loc='lower right')\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "_lEHXrzpP9cd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 927
        },
        "outputId": "4b0b0f67-80bc-45d1-b93f-4eeef5f6c0a1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "<font size='5'>Generator Loss</font>"
      ],
      "metadata": {
        "id": "017l0Bcs8EmC"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "with open(os.path.join(checkpoint_directory, \"gloss.pkl\"), 'rb') as f:\n",
        "  gloss_list = pickle.load(f)\n",
        "plt.plot(gloss_list)\n",
        "plt.title('Generator Loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend('g-loss', loc='lower right')\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "hfl-XuP6xPRc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 472
        },
        "outputId": "abbd7065-923b-42fc-a863-d5377a89b9b1"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Saving files on GitHub\n",
        "\n",
        "Configure git with mail, username and GitHub token"
      ],
      "metadata": {
        "id": "HrxGGL1rvgN2"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "!git config --global user.email $GIT_MAIL\n",
        "!git config --global user.name $GIT_USERNAME"
      ],
      "metadata": {
        "id": "pv3aAjyHvl1Z"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "%%bash\n",
        "cd /content/AE-WGAN-GP\n",
        "git remote set-url origin \"https://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@github.com/dado330/AE-WGAN-GP.git\""
      ],
      "metadata": {
        "id": "yUkysDCIvo99"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "Commit and push the changes to the repo"
      ],
      "metadata": {
        "id": "m-DPKCsFv723"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "%%bash\n",
        "cd /content/AE-WGAN-GP\n",
        "git add -A\n",
        "git commit -m \"Updated comments and generated data\"\n",
        "git push -u origin main"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TqhDsI5Kv3wR",
        "outputId": "dd1a42ac-c57c-49d5-8bbc-8d895fe0e020"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[main 0978597] Updated comments and generated data\n",
            " 3 files changed, 1 insertion(+), 1 deletion(-)\n",
            " rewrite AE_checkpoints/ckpt-1.data-00000-of-00001 (83%)\n",
            " rewrite AE_checkpoints/ckpt-1.index (93%)\n",
            "Branch 'main' set up to track remote branch 'main' from 'origin'.\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "To https://github.com/dado330/AE-WGAN-GP.git\n",
            "   e64e7c1..0978597  main -> main\n"
          ]
        }
      ]
    }
  ]
}