import argparse
import pickle
from model import NLG
from data_engine import DataEngine
from text_token import _UNK, _PAD, _BOS, _EOS
import torch
import torch.nn as nn
import numpy as np
import os
from utils import print_config, add_path
from model_utils import get_embeddings
from argument import define_arguments
from utils import get_time
_, args = define_arguments()
args = add_path(args)
if args.verbose_level > 0:
print_config(args)
use_cuda = torch.cuda.is_available()
train_data_engine = DataEngine(
data_dir=args.data_dir,
dataset=args.dataset,
save_path=args.train_data_file,
vocab_path=args.vocab_file,
is_spacy=args.is_spacy,
is_lemma=args.is_lemma,
fold_attr=args.fold_attr,
use_punct=args.use_punct,
vocab_size=args.vocab_size,
from argument import define_arguments
parser, args = define_arguments(script=True)
list_arg = set()
for arg in vars(args):
if type(getattr(args, arg)) == list:
if len(getattr(args, arg)) == 1:
setattr(args, arg, getattr(args, arg)[0])
else:
list_arg.add(arg)
loop_cnt = 0
for arg in vars(args):
if arg in list_arg:
loop_cnt += 1
attrs = getattr(args, arg)
print("{}for {} in {}; do".format(
"\t"*(loop_cnt-1),
arg,
' '.join(list(map(str, attrs)))))
print("{}python3 train.py \\".format("\t"*loop_cnt))
for arg in vars(args):
if arg in list_arg:
print("{}--{} {} \\".format("\t"*loop_cnt, arg, "${{{}}}".format(arg)))
else:
if getattr(args, arg) != parser.get_default(arg):
print("{}--{} {} \\".format(
"\t"*loop_cnt, arg, getattr(args, arg)))