bitnet5.py

import numpy as np
import pandas as pd
import tensorflow as tf
import keras
import datetime
from keras.layers import Concatenate
import csv
import hashlib
import secrets
import binascii
from binascii import unhexlify, hexlify

# load data from csv file into a pandas dataframe
df = pd.read_csv("train.csv")
id_data = df["id"].values
nonce_data = df["nonce"].values

#hash_data_0 = df["hash"].values
#for i in hash_data_0:
    #hash_data_1 = hash_data_0[i]
    #hash_data = int(hash_data_1, 16)

df["hash_int"] = df["hash"].apply(lambda x: int(x, 16))
hash_data = df["hash_int"].values
df["merkle_root_int"] = df["merkle_root"].apply(lambda x: int(x, 16))
merkle_root_data = df["merkle_root_int"].values


#time_data = df["time"].values
# Convert the string time value to a datetime object
#df["time_value"] = df["time"].datetime.datetime.strptime(time_data, '%m/%d/%y %H:%M')
# Convert the datetime object to a float value representing the number of seconds since the epoch
#time_data_float = (df["time_value"] - datetime.datetime(1970,1,1)).total_seconds()
#time_data = time_data_float.values
#print(time_data)

version_data = df["version"].values
bits_data = df["bits"].values

# normalize the input variable
# convert the input variable to a tensor
id_data_tensor = tf.constant(id_data, dtype=tf.float64)
nonce_data_tensor = tf.constant(nonce_data, dtype=tf.float64)
hash_data_tensor = tf.constant(hash_data, dtype=tf.float64)
merkle_root_data_tensor = tf.constant(merkle_root_data, dtype=tf.float64)
#time_data_tensor = tf.constant(time_data, dtype=tf.float64)
version_data_tensor = tf.constant(version_data, dtype=tf.float64)
bits_data = tf.constant(bits_data, dtype=tf.float64)

#X and Y variables
#Training Data
id_x = id_data_tensor[:-1]
nonce_x = nonce_data_tensor[:-1]
hash_x = hash_data_tensor[:-1] 
merkle_root_x = merkle_root_data_tensor[:-1]
#time_x = time_data_tensor[:-1]
version_x = version_data_tensor[:-1]
bits_x = bits_data[:-1] 

#Target Data
id_y = id_data_tensor[1:]
nonce_y = nonce_data_tensor[1:]
hash_y = hash_data_tensor[1:] 
merkle_root_y = merkle_root_data_tensor[1:]
#time_y = time_data_tensor[1:]
version_y = version_data_tensor[1:]
bits_y = bits_data[1:] 


print("XXXXXXXXXXXXXXX")
# create the model
model = tf.keras.models.Sequential([
    tf.keras.layers.Dense(5, activation="relu", input_shape=[1]),
    tf.keras.layers.Dense(25, activation="relu", name="layer1"),
    tf.keras.layers.Dense(25, activation="relu", name="layer2"),
    tf.keras.layers.Dense(1, activation="relu", name="layer3")
])

# compile the model
#model.compile(optimizer='adam', loss='mean_squared_error', learning_rate = 0.1)
optim = keras.optimizers.Adam(learning_rate=0.01)
#model.compile(optimizer=optim, loss='mean_squared_error', metrics=['accuracy'])
#model.compile(optimizer=optim, loss='huber_loss', metrics=['accuracy'])
# 1954562688.0000
model.compile(optimizer=optim, loss='huber_loss', metrics=['accuracy'])

# train the model on the data
history = model.fit(nonce_x, nonce_y, epochs=3)
#history = model.fit(x=[id_x, nonce_x, hash_x, merkle_root_x, version_x], y = nonce_y, epochs=2)

# make predictions using the trained model
predictions = model.predict(nonce_y)
print("Next nonce: ", predictions)

integer_list = [int(val[0]) for val in predictions]
with open('list.csv', 'w', newline='') as file:
    writer = csv.writer(file)
    writer.writerow(integer_list)