CosineSimilarity.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyNcphUaOOsxUBAGHmEFNZhw",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/ShuboLiu/MyColabSampless/blob/main/CosineSimilarity.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Insight of Cosine Similarity MAE"
      ],
      "metadata": {
        "id": "DeD0bnn3Lz-p"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "t1TgdOMcIpSK",
        "outputId": "8d950f20-11bf-4f57-ab9b-a1fb5ac8467b"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/gdrive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "\n",
        "drive.mount('/content/gdrive')"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!ls gdrive/MyDrive/temp/"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UZkvCTcgTmBm",
        "outputId": "f21a8910-a6f8-4466-a951-258926ba2945"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "eval.json\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "import json, os\n",
        "%matplotlib inline \n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "\n",
        "with open('/content/gdrive/MyDrive/temp/eval.json') as file:\n",
        "  content = json.load(file)\n",
        "with open('/content/gdrive/MyDrive/temp/eval_mae_notraining.json') as file:\n",
        "  content_MAE_notraning = json.load(file)\n",
        "\n",
        "similarity_unseen = []\n",
        "similarity_seen = []\n",
        "similarity_train = []\n",
        "loss_unseen = []\n",
        "loss_seen = []\n",
        "loss_train = []\n",
        "\n",
        "splits = ['train', 'val_seen', 'val_unseen']\n",
        "for split in splits:\n",
        "  similarity = []\n",
        "  loss = []\n",
        "  data = content['R2R_{}_enc'.format(split)]\n",
        "  for key in data.keys():\n",
        "    similarity.append(float(data[key]['cos_loss']))\n",
        "    loss.append(float(data[key]['l2_loss']))\n",
        "  \n",
        "  x = np.arange(len(similarity))*5\n",
        "  y_1 = similarity\n",
        "  y_2 = loss\n",
        "  \n",
        "  plt.plot(x, y_1, color='blue')\n",
        "  plt.plot(x, y_2, color='green')\n",
        "  plt.hlines(content_MAE_notraning['R2R_{}_enc'.format(split)]['NONE']['l2_loss'], xmin=min(x), xmax=max(x), color='green')\n",
        "  plt.hlines(content_MAE_notraning['R2R_{}_enc'.format(split)]['NONE']['cos_loss'], xmin=min(x), xmax=max(x), color='blue')\n",
        "\n",
        "  plt.title(\"{}\".format(split))\n",
        "  plt.show() \n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 809
        },
        "id": "ue_ouHvcQ7uL",
        "outputId": "a416ea42-1e04-4012-982c-1bb4b6c82e89"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "绿色：L2 Loss\n",
        "蓝色：Cosine Similarity\n",
        "横线：MAE开始训练前预测的结果与GT的对比【可当作Random Noise Baseline】\n",
        "\n",
        "这里可以看出Loss和Cosine Similarity的结果都是较为正常的：\n",
        "1. 在Train上的Loss呈现下降状态，而相比之下Cosine Similarity呈现上升效果。这体现模型具有较好的收敛效果\n",
        "2. 在Eval_seen结果上整体与Train上的结果保持一致。\n",
        "3. 与之形成鲜明对比的Val_unseen：Similarity先上升后逐渐下降。这反映了随着训练Epoch数量的增加，模型泛化性能逐步降低。\n",
        "\n",
        "但是为什么会在Epoch=5这么快的地方到达峰值值得进一步深思。造成这一现象的潜在原因包括：\n",
        "1. 模型太大了？相比于MAE的ImageNet图像输入而言，我们的Feature Mean输入数据量显著低，因此模型是不是也应该相对应地减小？\n",
        "2. 应当进一步测试 Epoch=15 附近在DUET模型中的预测效果"
      ],
      "metadata": {
        "id": "OKMWOtE0UitI"
      }
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "iiv_MZ1oT3rW"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}