mrq/DL-Art-School
1
3
Fork 5
Code Issues Pull Requests Packages Projects Releases Wiki Activity

mod trainer to be very explicit about the fact that loading models and state together dont work, but allow it

Browse Source
This commit is contained in:
James Betker 2021-10-28 22:32:42 -06:00
parent bb0a0c8264
commit 2afea126d7
2 changed files with 22 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
codes/train.py
View File

@ -43,7 +43,6 @@ class Trainer:
device_id = torch.cuda.current_device()
resume_state = torch.load(opt['path']['resume_state'],
map_location=lambda storage, loc: storage.cuda(device_id))
option.check_resume(opt, resume_state['iter']) # check resume options
else:
resume_state = None
@ -64,18 +63,15 @@ class Trainer:
# tensorboard logger
if opt['use_tb_logger'] and 'debug' not in opt['name']:
self.tb_logger_path = os.path.join(opt['path']['experiments_root'], 'tb_logger')
version = float(torch.__version__[0:3])
if version >= 1.1: # PyTorch 1.1
from torch.utils.tensorboard import SummaryWriter
else:
self.self.logger.info(
'You are using PyTorch {}. Tensorboard will use [tensorboardX]'.format(version))
from tensorboardX import SummaryWriter
self.tb_logger = SummaryWriter(log_dir=self.tb_logger_path)
else:
util.setup_logger('base', opt['path']['log'], 'train', level=logging.INFO, screen=True)
self.logger = logging.getLogger('base')
if resume_state is not None:
option.check_resume(opt, resume_state['iter']) # check resume options
# convert to NoneDict, which returns None for missing keys
opt = option.dict_to_nonedict(opt)
self.opt = opt
@ -284,7 +280,7 @@ class Trainer:
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument('-opt', type=str, help='Path to option YAML file.', default='../options/train_gpt_asr_mass_distill.yml')
parser.add_argument('-opt', type=str, help='Path to option YAML file.', default='../options/train_noisy_audio_clips_classifier.yml')
parser.add_argument('--launcher', choices=['none', 'pytorch'], default='none', help='job launcher')
parser.add_argument('--local_rank', type=int, default=0)
args = parser.parse_args()

15
codes/utils/options.py
View File

@ -103,7 +103,20 @@ def check_resume(opt, resume_iter):
# Automatically fill in the network paths for a given resume iteration.
for k in opt['networks'].keys():
pt_key = 'pretrain_model_%s' % (k,)
assert pt_key not in opt['path'].keys() # There's no real reason to load from a training_state AND a model.
if pt_key in opt['path'].keys():
# This is a dicey, error prone situation that has bitten me in both ways it can be handled. Opt for
# a big, verbose error message.
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
print("!!!!!!!!!!!!!!WARNING!! YOU SPECIFIED A PRETRAINED MODEL PATH!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
print("!!!!!!!!!!!!!!!!!!!!!!!!AND A RESUME STATE PATH. THERE IS NO!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
print("!!!!!!!!!!!!!!!!!!!!!!!!GOOD WAY TO HANDLE THIS SO WE JUST IGNORE!!!!!!!!!!!!!!!!!!!!!!!!!!!")
print("!!!!!!!!!!!!!!!!!!!!!!!!THE MODEL PATH!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
print("!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!")
else:
opt['path'][pt_key] = osp.join(opt['path']['models'],
'{}_{}.pth'.format(resume_iter, k))
logger.info('Set model [%s] to %s' % (k, opt['path'][pt_key]))