cfeng783 / gtt Goto Github PK

Dear developer. I would like to draw your attention to the fact that at the moment, finetuning on datasets with the purpose of univariate forecast, is worse in terms of forecast accuracy than zero-shot forecasting

please give download link

Thanks for your work, I have some doubts, the small version of the model without fine-tuning performed very poorly on my dataset, so I tried to change the model with fine-tuning, first I tried fine-tuning on the official electricity dataset, fine-tuning After success, I encountered the following error when using it.

{'ret': 'fail', 'message': 'all the input array dimensions except for the concatenation axis must match exactly, but along dimension 1, the array at index 0 has size 720 and the array at index 1 has size 64'}

I would be grateful if you could provide an API for the fine-tuned model.