text
stringlengths
1
93.6k
# Remaining keys will be printed "as is"
# (usually in alphabet order)
result.update(**data)
return dumps(result, default=str)
processors = list()
# In some cases there is no need to print a timestamp,
# because it is already added by an upstream service, such as systemd
if log_config.show_datetime is True:
processors.append(structlog.processors.TimeStamper(
fmt=log_config.datetime_format,
utc=log_config.time_in_utc
)
)
# Always add a log level
processors.append(structlog.processors.add_log_level)
# Render selection: JSON or for output to terminal
if log_config.renderer == LogRenderer.JSON:
processors.append(structlog.processors.JSONRenderer(serializer=custom_json_serializer))
else:
processors.append(structlog.dev.ConsoleRenderer(
# You can turn off colors in the logs
colors=log_config.use_colors_in_console,
# You can remove padding in levels, i.e. instead of
# [info ] Some info log
# [warning] Some warning log
# will be
# [info] Some info log
# [warning] Some warning log
pad_level=True
))
return processors
# <FILESEP>
#!/usr/bin/env python3 -u
# Copyright (c) Facebook, Inc. and its affiliates.
#
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
"""
Train a new model on one or across multiple GPUs.
"""
import collections
import math
import random
import numpy as np
import torch
from fairseq import checkpoint_utils, distributed_utils, options, progress_bar, tasks, utils
from fairseq.data import iterators
from fairseq.trainer import Trainer
from fairseq.meters import AverageMeter, StopwatchMeter
def main(args, init_distributed=False):
utils.import_user_module(args)
assert args.max_tokens is not None or args.max_sentences is not None, \
'Must specify batch size either with --max-tokens or --max-sentences'
# Initialize CUDA and distributed training
if torch.cuda.is_available() and not args.cpu:
torch.cuda.set_device(args.device_id)
torch.manual_seed(args.seed)
if init_distributed:
args.distributed_rank = distributed_utils.distributed_init(args)
if distributed_utils.is_master(args):
checkpoint_utils.verify_checkpoint_directory(args.save_dir)
#Print args
#print(args)
# Setup task, e.g., translation, language modeling, etc.
#The loading of the text dictionary thing happens in def setup_task in the file sentence_preciction.py
task = tasks.setup_task(args)
# Load valid dataset (we load training data below, based on the latest checkpoint)
for valid_sub_split in args.valid_subset.split(','):
task.load_dataset(valid_sub_split, combine=False, epoch=0)
# Build model and criterion
model = task.build_model(args)
criterion = task.build_criterion(args)
#print(model)
print('| model {}, criterion {}'.format(args.arch, criterion.__class__.__name__))