pytorch-lightning
pytorch-lightning copied to clipboard
Lightning-AI
neptune.ai logger produces lots of errors when logging "training/epoch"
Bug description
Neptune logger gives a lot of errors like "[neptune] [error ] Error occurred during asynchronous operation processing: X-coordinates (step) must be strictly increasing for series attribute: training/epoch. Invalid point: 34.0"
Those are actually false positives, the "training/epoch" curve in the neptune UI looks fine.
similar to https://github.com/Lightning-AI/pytorch-lightning/issues/2946
What version are you seeing the problem on?
v2.2
How to reproduce the bug
setup NEPTUNE_API_TOKEN and NEPTUNE_PROJECT first for a proper connection to neptune.ai
import os
import lightning as lit
import torch
from lightning.pytorch.loggers import NeptuneLogger
from torch.utils.data import Dataset, DataLoader
class DummyDataset(Dataset):
def __init__(self):
pass
def __len__(self):
return 100
def __getitem__(self, item):
return {"image": torch.rand(3, 16, 16), "label": torch.randint(0, 100, (1,))}
class DummyModel(lit.LightningModule):
def __init__(self):
super().__init__()
self.model = torch.nn.Linear(3 * 16 * 16, 100)
self.epoch_identifier = "dummy"
def forward(self, x):
return self.model(x)
def training_step(self, batch, batch_idx):
x, y = batch["image"], batch["label"]
x = x.view(x.size(0), -1)
y = y.view(-1)
logits = self.model(x)
loss = torch.nn.functional.cross_entropy(logits, y)
self.log("train_loss", loss, on_step=True, on_epoch=True, prog_bar=True, logger=True)
return loss
def configure_optimizers(self):
optimizer = torch.optim.Adam(self.parameters(), lr=1e-3)
return optimizer
def validation_step(self, batch, batch_idx):
x, y = batch["image"], batch["label"]
x = x.view(x.size(0), -1)
y = y.view(-1)
logits = self.model(x)
loss = torch.nn.functional.cross_entropy(logits, y)
self.log("val_loss", loss, on_step=True, on_epoch=True, prog_bar=True, logger=True)
def test_step(self, batch, batch_idx):
return self.validation_step(batch, batch_idx)
def main():
wlogger = NeptuneLogger(log_model_checkpoints=False)
output_dir = "temp_lit"
os.makedirs(output_dir, exist_ok=True)
trainer = lit.Trainer(
devices=1,
default_root_dir=output_dir,
logger=wlogger,
max_epochs=5,
enable_progress_bar=False,
log_every_n_steps=5,
)
model = DummyModel()
dataset = DummyDataset()
train_loader = DataLoader(dataset, batch_size=16, num_workers=4)
val_loader = DataLoader(dataset, batch_size=16, num_workers=4)
trainer.fit(model, train_loader, val_loader)
if __name__ == "__main__":
main()
### Error messages and logs
GPU available: True (cuda), used: True TPU available: False, using: 0 TPU cores IPU available: False, using: 0 IPUs HPU available: False, using: 0 HPUs [neptune] [info ] Neptune initialized. Open in the app: https://app.neptune.ai/ [...] LOCAL_RANK: 0 - CUDA_VISIBLE_DEVICES: [0,1]
| Name | Type | Params
0 | model | Linear | 76.9 K
76.9 K Trainable params
0 Non-trainable params
76.9 K Total params
0.308 Total estimated model params size (MB)
[neptune] [error ] Error occurred during asynchronous operation processing: X-coordinates (step) must be strictly increasing for series attribute: training/epoch. Invalid point: 6.0
[neptune] [error ] Error occurred during asynchronous operation processing: X-coordinates (step) must be strictly increasing for series attribute: training/epoch. Invalid point: 13.0
[neptune] [error ] Error occurred during asynchronous operation processing: X-coordinates (step) must be strictly increasing for series attribute: training/epoch. Invalid point: 20.0
[neptune] [error ] Error occurred during asynchronous operation processing: X-coordinates (step) must be strictly increasing for series attribute: training/epoch. Invalid point: 27.0
Trainer.fit stopped: max_epochs=5 reached.
[neptune] [info ] Shutting down background jobs, please wait a moment...
[neptune] [info ] Done!
[neptune] [info ] Waiting for the remaining 17 operations to synchronize with Neptune. Do not kill this process.
[neptune] [error ] Error occurred during asynchronous operation processing: X-coordinates (step) must be strictly increasing for series attribute: training/epoch. Invalid point: 34.0
[neptune] [error ] Error occurred during asynchronous operation processing: X-coordinates (step) must be strictly increasing for series attribute: training/epoch. Invalid point: 34.0
[neptune] [info ] All 17 operations synced, thanks for waiting!
[neptune] [info ] Explore the metadata in the Neptune app: https://app.neptune.ai/ [...]
### Environment
<details>
<summary>Current environment</summary>
* CUDA:
- GPU:
- NVIDIA GeForce GTX 1070 Ti
- Quadro P400
- available: True
- version: 12.1
* Lightning:
- lightning: 2.2.1
- lightning-utilities: 0.11.0
- pytorch-lightning: 2.2.1
- torch: 2.2.1
- torchaudio: 2.2.1
- torchmetrics: 1.3.2
- torchvision: 0.17.1
* Packages:
- aiohttp: 3.9.3
- aiosignal: 1.3.1
- arrow: 1.3.0
- async-timeout: 4.0.3
- attrs: 23.2.0
- boto3: 1.34.66
- botocore: 1.34.66
- bravado: 11.0.3
- bravado-core: 6.1.1
- brotli: 1.0.9
- certifi: 2024.2.2
- charset-normalizer: 2.0.4
- click: 8.1.7
- filelock: 3.13.1
- fqdn: 1.5.1
- frozenlist: 1.4.1
- fsspec: 2024.3.1
- future: 1.0.0
- gitdb: 4.0.11
- gitpython: 3.1.42
- gmpy2: 2.1.2
- idna: 3.4
- isoduration: 20.11.0
- jinja2: 3.1.3
- jmespath: 1.0.1
- jsonpointer: 2.4
- jsonref: 1.1.0
- jsonschema: 4.21.1
- jsonschema-specifications: 2023.12.1
- lightning: 2.2.1
- lightning-utilities: 0.11.0
- markupsafe: 2.1.3
- mkl-fft: 1.3.8
- mkl-random: 1.2.4
- mkl-service: 2.4.0
- monotonic: 1.6
- mpmath: 1.3.0
- msgpack: 1.0.8
- multidict: 6.0.5
- neptune: 1.9.1
- networkx: 3.1
- numpy: 1.26.4
- oauthlib: 3.2.2
- packaging: 24.0
- pandas: 2.2.1
- pillow: 10.2.0
- pip: 23.3.1
- psutil: 5.9.8
- pyjwt: 2.8.0
- pysocks: 1.7.1
- python-dateutil: 2.9.0.post0
- pytorch-lightning: 2.2.1
- pytz: 2024.1
- pyyaml: 6.0.1
- referencing: 0.34.0
- requests: 2.31.0
- requests-oauthlib: 1.4.0
- rfc3339-validator: 0.1.4
- rfc3986-validator: 0.1.1
- rpds-py: 0.18.0
- s3transfer: 0.10.1
- setuptools: 68.2.2
- simplejson: 3.19.2
- six: 1.16.0
- smmap: 5.0.1
- swagger-spec-validator: 3.0.3
- sympy: 1.12
- torch: 2.2.1
- torchaudio: 2.2.1
- torchmetrics: 1.3.2
- torchvision: 0.17.1
- tqdm: 4.66.2
- triton: 2.2.0
- types-python-dateutil: 2.9.0.20240316
- typing-extensions: 4.9.0
- tzdata: 2024.1
- uri-template: 1.3.0
- urllib3: 2.1.0
- webcolors: 1.13
- websocket-client: 1.7.0
- wheel: 0.41.2
- yarl: 1.9.4
* System:
- OS: Linux
- architecture:
- 64bit
- ELF
- processor: x86_64
- python: 3.10.13
- release: 5.4.0-172-generic
- version: #190-Ubuntu SMP Fri Feb 2 23:24:22 UTC 2024
</details>
### More info
_No response_
You can follow updates to this issue here: https://github.com/neptune-ai/neptune-client/issues/1702
Be assured, though, that this "error" does not lead to data loss. It is merely caused by the training and validation loops trying to log the epoch number to the same namespace. Since it is already logged, and Neptune always expects the "step" to be increasing, the duplicate epoch value at the same step is dropped.
More details about this error here: https://docs.neptune.ai/help/error_step_must_be_increasing/
