search

SalesforceAIResearch / uni2ts

[ICML2024] Unified Training of Universal Time Series Forecasting Transformers
Apache License 2.0
611 stars 49 forks source link

Finetuning bug #38

Open wyhzunzun123123 opened 1 month ago

wyhzunzun123123 commented 1 month ago

Hello, I really appreciate your work as it got such a surprising performance in zero-shot setting. But I‘ve gotten a problem while fine-tuning using your command 'python -m cli.finetune run_name=example_run model=moirai_1.0_R_small data=etth1 val_data=etth1'. The specific bug is 'UserWarning: resource_tracker: There appear to be 12 leaked semaphore objects to clean up at shutdown warnings.warn('resource_tracker: There appear to be %d ''. Although I tried so many methods provided by the internet, this issue still can't be addressed. Could you give some advice?

gorold commented 1 month ago

I think you can ignore this, seems like some issue with the dataloading.