def main():
args = get_args()
with open(args.config_path) as f:
hparams = yaml.load(f, Loader=yaml.SafeLoader)
neptune_logger = NeptuneLogger(
api_key=os.environ["NEPTUNE_API_TOKEN"],
project_name=hparams["project_name"],
experiment_name=f"{hparams['experiment_name']}", # Optional,
tags=["pytorch-lightning", "mlp"], # Optional,
upload_source_files=[],
)
pipeline = RetinaFaceMask(hparams)
Path(hparams["checkpoint_callback"]["filepath"]).mkdir(exist_ok=True, parents=True)
trainer = object_from_dict(
hparams["trainer"],
logger=neptune_logger,
checkpoint_callback=object_from_dict(hparams["checkpoint_callback"]),
)
trainer.fit(pipeline)
def test_neptune_pickle(tmpdir):
"""Verify that pickling trainer with neptune logger works."""
tutils.reset_seed()
from pytorch_lightning.logging import NeptuneLogger
# hparams = tutils.get_hparams()
# model = LightningTestModel(hparams)
logger = NeptuneLogger(offline_mode=True)
trainer_options = dict(default_save_path=tmpdir,
max_epochs=1,
logger=logger)
trainer = Trainer(**trainer_options)
pkl_bytes = pickle.dumps(trainer)
trainer2 = pickle.loads(pkl_bytes)
trainer2.logger.log_metrics({"acc": 1.0})
def test_neptune_logger(tmpdir):
"""Verify that basic functionality of neptune logger works."""
tutils.reset_seed()
from pytorch_lightning.logging import NeptuneLogger
hparams = tutils.get_hparams()
model = LightningTestModel(hparams)
logger = NeptuneLogger(offline_mode=True)
trainer_options = dict(default_save_path=tmpdir,
max_epochs=1,
train_percent_check=0.01,
logger=logger)
trainer = Trainer(**trainer_options)
result = trainer.fit(model)
print('result finished')
assert result == 1, "Training failed"
def configure_optimizers(self):
return torch.optim.Adam(self.parameters())
def train_dataloader(self):
self.dset = MathDataset("data/raw/math.train")
train_loader = DataLoader(
self.dset, batch_size=64, shuffle=True, collate_fn=self.dset.collate,
)
return train_loader
def val_dataloader(self):
dset = MathDataset("data/raw/math.val")
self.val_dset = dset
loader = DataLoader(
dset, batch_size=64, shuffle=False, collate_fn=dset.collate,
)
return loader
if __name__ == "__main__":
model = SimpleSeq2Seq()
trainer = pl.Trainer(
gpus=[0],
gradient_clip_val=1.0,
max_epochs=50,
fast_dev_run=False,
logger=NeptuneLogger(project_name="mbednarski/seq2seq"),
)
trainer.fit(model)