from utils.testing import run_tests, compile_programs
from utils.tests import CombingTest
# CXX_COMPILER_PATH = '/usr/bin/g++-10'
CXX_COMPILER_PATH = '/usr/bin/g++'
SINGLE_THREADED_SOLUTIONS = []
MULTI_THREADED_SOLUTIONS = [
# 'semi_local_hybrid_iterative'
'semi_local_parallel_iterative'
]
SOLUTIONS_FOLDER = 'combing_solutions' # where we put our ready to run implementations
default_logger = Logger('fig9_log.txt')
build_logger = Logger('fig9_build_log.txt')
def boolean_string(s):
if s not in {'False', 'True'}:
raise ValueError('Not a valid boolean string')
return s == 'True'
if __name__ == '__main__':
arg_parser = ArgumentParser()
arg_parser.add_argument('max_thds', type=int, help='Threads')
arg_parser.add_argument(
'tests',
type=str,
#!/usr/bin/python3
# -*- coding: utf-8 -*-
# @author: sharly
import configparser
import yaml
import openpyxl
import csv
import time
from utils.logger import Logger
logger = Logger(logger="BaseUtil").get_log()
class BaseUtil:
def __init__(self):
self.root_path = 'E:/PycharmProjects/wechat_test'
def set_root_path(self, path):
"""设置项目的主path"""
self.root_path = path
def get_root_path(self):
"""获取项目的主path"""
return self.root_path
def get_config_value(self, section, key):
"""获取config文件中section的key值"""
config_path = self.root_path + "/conf/config.ini"
config = configparser.ConfigParser()
def main(args):
# args.n_epochs = 10
# args.crop_scale = 0.3
# args.batch_size = 128
args.normal_data_ratio = 0.9
if args.dataset == 'MNIST':
args.sample_size = 28
elif args.dataset == 'SOP' or 'Shopee':
if args.model == 'resnet':
args.sample_size = 224
elif args.model == 'vgg' or args.model == 'vgg_attn':
args.sample_size = 224
elif args.model == 'inception':
args.sample_size = 299
else:
args.sample_size = 224
spatial_transform_train = get_train_transform(args)
crop_transform = get_crop_transform(args)
if args.dataset == 'MNIST':
train_data_loader = CroppedMNISTLoader(args, crop_transform=crop_transform,
spatial_transform=spatial_transform_train, training=True)
elif args.dataset == 'SOP':
train_data_loader = CroppedSOPLoader(args, crop_transform=crop_transform,
spatial_transform=spatial_transform_train, training=True)
elif args.dataset == 'Shopee':
train_data_loader = ShopeeDataLoader(args, crop_transform=crop_transform,
spatial_transform=spatial_transform_train, training=True)
valid_data_loader = train_data_loader.split_validation()
args.n_channels = train_data_loader.n_channels
args.n_classes = train_data_loader.n_classes
model, parameters = get_model(args)
model = model.to(device)
criterion = losses.cross_entropy_loss()
train_logger = Logger(
os.path.join(args.log_path, 'train.log'),
['epoch', 'loss', 'acc', 'lr'])
train_batch_logger = Logger(
os.path.join(args.log_path, 'train_batch.log'),
['epoch', 'batch', 'iter', 'loss', 'acc', 'lr'])
valid_logger = Logger(
os.path.join(args.log_path, 'val.log'),
['epoch', 'loss', 'acc'])
revision_logger = Logger(
os.path.join(args.log_path, 'revision_info.log'),
['dataset', 'dataset_size', 'train_test_split', 'model', 'model_depth', 'resume', 'resume_path', 'batch_size',
'n_epochs', 'sample_size', 'crop_scale', 'crop_transform', 'cropped_data_ratio'])
revision_logger.log({
'dataset': args.dataset,
'dataset_size': args.dataset_size,
'train_test_split': args.train_test_split,
'model': args.model,
'model_depth': args.model_depth,
'resume': args.resume,
'resume_path': args.resume_path,
'batch_size': args.batch_size,
'n_epochs': args.n_epochs,
'sample_size': args.sample_size,
'crop_scale': args.crop_scale,
'crop_transform': crop_transform.__class__.__name__,
'cropped_data_ratio': args.cropped_data_ratio
})
if args.nesterov:
dampening = 0
else:
dampening = args.dampening
optimizer = optim.SGD(
parameters,
lr=args.learning_rate,
momentum=args.momentum,
dampening=dampening,
weight_decay=args.weight_decay,
nesterov=args.nesterov)
scheduler = lr_scheduler.ReduceLROnPlateau(
optimizer, 'min', patience=args.lr_patience)
trainer = Trainer(model, criterion, optimizer, args, device, train_data_loader, lr_scheduler=scheduler,
valid_data_loader=valid_data_loader, train_logger=train_logger, batch_logger=train_batch_logger, valid_logger=valid_logger)
trainer.train()
