示例#1
0
import numpy as np
from math import sqrt
from sklearn.metrics import r2_score
from sklearn.metrics import mean_squared_error
from sklearn.metrics import accuracy_score
from sklearn import metrics
from model_OCD import OCD

import checkmate as cm
import data_helpers as dh

# Parameters
# ==================================================

logger = dh.logger_fn(
    "tflog",
    "logs/training_kfold_{0}_{1}_time_{2}.log".format(sys.argv[1], sys.argv[0],
                                                      int(time.time())))

kfold = int(sys.argv[1])
batch_size = int(sys.argv[2])

tf.compat.v1.flags.DEFINE_float("learning_rate", 0.001, "Learning rate")
tf.compat.v1.flags.DEFINE_float("keep_prob", 0.5,
                                "Keep probability for dropout")
tf.compat.v1.flags.DEFINE_integer("hidden_size", 256,
                                  "The number of hidden nodes (Integer)")
tf.compat.v1.flags.DEFINE_integer("evaluation_interval", 1,
                                  "Evaluate and print results every x epochs")
tf.compat.v1.flags.DEFINE_integer("batch_size", batch_size,
                                  "Batch size for training.")
tf.compat.v1.flags.DEFINE_integer("epochs", 6,
示例#2
0
while not (TRAIN_OR_RESTORE.isalpha() and TRAIN_OR_RESTORE.upper() in ['T', 'R']):
    TRAIN_OR_RESTORE = input('✘ The format of your input is illegal, please re-input: ')
logging.info('✔︎ The format of your input is legal, now loading to next step...')

TRAIN_OR_RESTORE = TRAIN_OR_RESTORE.upper()

CLASS_BIND = input("☛ Use Class Bind or Not?(Y/N) \n")
while not (CLASS_BIND.isalpha() and CLASS_BIND.upper() in ['Y', 'N']):
    CLASS_BIND = input('✘ The format of your input is illegal, please re-input: ')
logging.info('✔︎ The format of your input is legal, now loading to next step...')

CLASS_BIND = CLASS_BIND.upper()

if TRAIN_OR_RESTORE == 'T':
    logger = data_helpers.logger_fn('tflog', 'training-{}.log'.format(time.asctime()))
if TRAIN_OR_RESTORE == 'R':
    logger = data_helpers.logger_fn('tflog', 'restore-{}.log'.format(time.asctime()))

TRAININGSET_DIR = '../Train.json'
VALIDATIONSET_DIR = '../Validation_bind.json'

# Data loading params
tf.flags.DEFINE_string("training_data_file", TRAININGSET_DIR, "Data source for the training data.")
tf.flags.DEFINE_string("validation_data_file", VALIDATIONSET_DIR, "Data source for the validation data.")
tf.flags.DEFINE_string("train_or_restore", TRAIN_OR_RESTORE, "Train or Restore.")
tf.flags.DEFINE_string("use_classbind_or_not", CLASS_BIND, "Use the class bind info or not.")

# Model Hyperparameterss
tf.flags.DEFINE_float("learning_rate", 0.001, "The learning rate (default: 0.001)")
tf.flags.DEFINE_integer("pad_seq_len", 150, "Recommand padding Sequence length of data (depends on the data)")
示例#3
0
# -*- coding:utf-8 -*-

import os
import sys
import time
import tensorflow as tf
import data_helpers as dh

# Parameters
# ==================================================

logger = dh.logger_fn('tflog', 'logs/test-{0}.log'.format(time.asctime()))

# Data Parameters
tf.flags.DEFINE_string("training_data_file", "./data/train_data_set.txt",
                       "Data source for the training data.")
tf.flags.DEFINE_string("validation_data_file", "./data/val_data_set.txt",
                       "Data source for the validation data")
tf.flags.DEFINE_string("test_data_file", "./data/test_data_set.txt",
                       "Data source for the test data")
tf.flags.DEFINE_string("checkpoint_dir", "./",
                       "Checkpoint directory from training run")
# tf.flags.DEFINE_string("vocab_data_file", "./", "Vocabulary file")

# Model Hyperparameters
# tf.flags.DEFINE_integer("pad_seq_len", 100, "Recommended padding Sequence length of data (depends on the data)")
tf.flags.DEFINE_integer(
    "embedding_dim", 128,
    "Dimensionality of character embedding (default: 128)")
tf.flags.DEFINE_integer("embedding_type", 1, "The embedding type (default: 1)")
tf.flags.DEFINE_integer(
示例#4
0
# -*- coding:utf-8 -*-

import os
import time
import numpy as np
import tensorflow as tf
import data_helpers

# Parameters
# ==================================================

logger = data_helpers.logger_fn('tflog', 'test-{}.log'.format(time.asctime()))

MODEL = input(
    "☛ Please input the model file you want to test, it should be like(1490175368): "
)

while not (MODEL.isdigit() and len(MODEL) == 10):
    MODEL = input(
        '✘ The format of your input is illegal, it should be like(1490175368), please re-input: '
    )
logger.info(
    '✔︎ The format of your input is legal, now loading to next step...')

CLASS_BIND = input("☛ Use Class Bind or Not?(Y/N) \n")
while not (CLASS_BIND.isalpha() and CLASS_BIND.upper() in ['Y', 'N']):
    CLASS_BIND = input(
        '✘ The format of your input is illegal, please re-input: ')
logger.info(
    '✔︎ The format of your input is legal, now loading to next step...')
示例#5
0
from batch_loader import BatchLoader, MulBatchLoader

# Parameters
# ==================================================

# TRAIN_OR_RESTORE = input("☛ Train or Restore?(T/R) \n")
TRAIN_OR_RESTORE = "T"

while not (TRAIN_OR_RESTORE.isalpha() and TRAIN_OR_RESTORE.upper() in ['T', 'R']):
    TRAIN_OR_RESTORE = input('✘ The format of your input is illegal, please re-input: ')
logging.info('✔︎ The format of your input is legal, now loading to next step...')

TRAIN_OR_RESTORE = TRAIN_OR_RESTORE.upper()

if TRAIN_OR_RESTORE == 'T':
    logger = dh.logger_fn('tflog', 'logs/training-{0}.log'.format(time.asctime()))
if TRAIN_OR_RESTORE == 'R':
    logger = dh.logger_fn('tflog', 'logs/restore-{0}.log'.format(time.asctime()))

TRAININGSET_DIR = '../data/Train.json'
VALIDATIONSET_DIR = '../data/Validation.json'
METADATA_DIR = '../data/metadata.tsv'


parser = argparse.ArgumentParser(description="Training CNN")
parser.add_argument("--training_data_file", type=str, default=TRAININGSET_DIR, help="Data source for the training data.")
parser.add_argument("--validation_data_file", type=str, default=VALIDATIONSET_DIR, help="Data source for the validation data.")
parser.add_argument("--metadata_file", type=str, default=METADATA_DIR, help="Metadata file for embedding visualization(Each line is a word segment in metadata_file).")
parser.add_argument("--train_or_restore", type=str, default=TRAIN_OR_RESTORE, help="Train or Restore.")
parser.add_argument("--data_path", type=str, default="data/train_snli.txt", help="Data path.")
parser.add_argument("--pad_seq_len", type=int, default=120, help="Recommended padding Sequence length of data (depends on the data)")
示例#6
0
import data_helpers as dh


# Parameters
# ==================================================

TRAIN_OR_RESTORE = 'T' #input("Train or Restore?(T/R): ")

while not (TRAIN_OR_RESTORE.isalpha() and TRAIN_OR_RESTORE.upper() in ['T', 'R']):
    TRAIN_OR_RESTORE = input("The format of your input is illegal, please re-input: ")
logging.info("The format of your input is legal, now loading to next step...")

TRAIN_OR_RESTORE = TRAIN_OR_RESTORE.upper()

if TRAIN_OR_RESTORE == 'T':
    logger = dh.logger_fn("tflog", "logs/training_kfold_{0}_{1}_time_{2}.log".format(sys.argv[1], sys.argv[0], int(time.time())))
if TRAIN_OR_RESTORE == 'R':
    logger = dh.logger_fn("tflog", "logs/restore-{0}.log".format(time.asctime()).replace(':', '_'))

kfold= int(sys.argv[1])
batch_size = int(sys.argv[2])
tf.compat.v1.flags.DEFINE_string("train_or_restore", TRAIN_OR_RESTORE, "Train or Restore.")
tf.compat.v1.flags.DEFINE_float("learning_rate", 0.001, "Learning rate")
tf.compat.v1.flags.DEFINE_float("norm_ratio", 5, "The ratio of the sum of gradients norms of trainable variable (default: 1.25)")
tf.compat.v1.flags.DEFINE_float("keep_prob", 0.5, "Keep probability for dropout")
tf.compat.v1.flags.DEFINE_float("radio", 0.6, "split radio")
tf.compat.v1.flags.DEFINE_integer("hidden_size", 256, "The number of hidden nodes (Integer)")
tf.compat.v1.flags.DEFINE_integer("evaluation_interval", 1, "Evaluate and print results every x epochs")
tf.compat.v1.flags.DEFINE_integer("batch_size", batch_size , "Batch size for training.")
tf.compat.v1.flags.DEFINE_integer("epochs", 3, "Number of epochs to train for.")
tf.compat.v1.flags.DEFINE_integer("kfold", kfold, "Number of epochs to train for.")
示例#7
0
import numpy as np
import pandas as pd
import tensorflow as tf
from sklearn import metrics
from math import sqrt
from sklearn.metrics import r2_score
from sklearn.metrics import mean_squared_error
import checkmate as cmm
import data_helpers as dh
import json

# Parameters
# ==================================================
#seq_len= int(sys.argv[1])
#batch_size = int(sys.argv[2])
logger = dh.logger_fn(
    "tflog", "logs/test-{0}.log".format(time.asctime()).replace(':', '_'))
number = sys.argv[1]
file_name = sys.argv[2]

MODEL = file_name
while not (MODEL.isdigit() and len(MODEL) == 10):
    MODEL = input(
        "The format of your input is illegal, it should be like(90175368), please re-input: "
    )
logger.info("The format of your input is legal, now loading to next step...")

TESTSET_DIR = 'data/assist2009_updated_all.csv'
MODEL_DIR = 'runs/' + MODEL + '/checkpoints/'
BEST_MODEL_DIR = 'runs/' + MODEL + '/bestcheckpoints/'
SAVE_DIR = 'results/' + MODEL