(
self,
logger=True,
checkpoint_callback=True,
default_save_path=None,
gradient_clip_val=0,
process_position=0,
gpus=-1,
log_gpu_memory=None,
show_progress_bar=True,
track_grad_norm=-1,
check_val_every_n_epoch=1,
accumulate_grad_batches=1,
max_updates=1000,
min_epochs=1,
val_check_interval=1.0,
log_save_interval=100,
row_log_interval=10,
print_nan_grads=False,
weights_summary='full',
num_sanity_val_steps=5,
resume_from_checkpoint=None,
)
| 353 | |
| 354 | class BaseTrainer: |
| 355 | def __init__( |
| 356 | self, |
| 357 | logger=True, |
| 358 | checkpoint_callback=True, |
| 359 | default_save_path=None, |
| 360 | gradient_clip_val=0, |
| 361 | process_position=0, |
| 362 | gpus=-1, |
| 363 | log_gpu_memory=None, |
| 364 | show_progress_bar=True, |
| 365 | track_grad_norm=-1, |
| 366 | check_val_every_n_epoch=1, |
| 367 | accumulate_grad_batches=1, |
| 368 | max_updates=1000, |
| 369 | min_epochs=1, |
| 370 | val_check_interval=1.0, |
| 371 | log_save_interval=100, |
| 372 | row_log_interval=10, |
| 373 | print_nan_grads=False, |
| 374 | weights_summary='full', |
| 375 | num_sanity_val_steps=5, |
| 376 | resume_from_checkpoint=None, |
| 377 | ): |
| 378 | self.log_gpu_memory = log_gpu_memory |
| 379 | self.gradient_clip_val = gradient_clip_val |
| 380 | self.check_val_every_n_epoch = check_val_every_n_epoch |
| 381 | self.track_grad_norm = track_grad_norm |
| 382 | self.on_gpu = True if (gpus and torch.cuda.is_available()) else False |
| 383 | self.process_position = process_position |
| 384 | self.weights_summary = weights_summary |
| 385 | self.max_updates = max_updates |
| 386 | self.min_epochs = min_epochs |
| 387 | self.num_sanity_val_steps = num_sanity_val_steps |
| 388 | self.print_nan_grads = print_nan_grads |
| 389 | self.resume_from_checkpoint = resume_from_checkpoint |
| 390 | self.default_save_path = default_save_path |
| 391 | |
| 392 | # training bookeeping |
| 393 | self.total_batch_idx = 0 |
| 394 | self.running_loss = [] |
| 395 | self.avg_loss = 0 |
| 396 | self.batch_idx = 0 |
| 397 | self.tqdm_metrics = {} |
| 398 | self.callback_metrics = {} |
| 399 | self.num_val_batches = 0 |
| 400 | self.num_training_batches = 0 |
| 401 | self.num_test_batches = 0 |
| 402 | self.get_train_dataloader = None |
| 403 | self.get_test_dataloaders = None |
| 404 | self.get_val_dataloaders = None |
| 405 | self.is_iterable_train_dataloader = False |
| 406 | |
| 407 | # training state |
| 408 | self.model = None |
| 409 | self.testing = False |
| 410 | self.disable_validation = False |
| 411 | self.lr_schedulers = [] |
| 412 | self.optimizers = None |
no test coverage detected