TimesNet for imputation ERROR Exception: The size of tensor a (5461) must match the size of tensor b (1000) at non-singleton dimension 1

[xiaoenli4816]

Issue description

The TimesNet method was used to recover the missing value of long time series observation meteorological data (11 variables).
The error is as follows:
2024-06-07 12:58:25 [INFO]: No given device, using default device: cpu
2024-06-07 12:58:25 [INFO]: Model files will be saved to imputation/timesnet\20240607_T125825
2024-06-07 12:58:25 [INFO]: Tensorboard file will be saved to imputation/timesnet\20240607_T125825\tensorboard
2024-06-07 12:58:25 [INFO]: TimesNet initialized with the given hyperparameters, the number of trainable parameters: 5,414,155
2024-06-07 12:58:25 [ERROR]: ❌ Exception: The size of tensor a (5461) must match the size of tensor b (1000) at non-singleton dimension 1
The length of the dataset is 5461 rows of records.

[github-actions]

Hi there 👋,

Thank you so much for your attention to PyPOTS! You can follow me on GitHub to receive the latest news of PyPOTS. If you find PyPOTS helpful to your work, please star⭐️ this repository. Your star is your recognition, which can help more people notice PyPOTS and grow PyPOTS community. It matters and is definitely a kind of contribution to the community.

I have received your message and will respond ASAP. Thank you for your patience! 😃

Best,
Wenjie

[WenjieDu]

This issue is caused by the overlong length that 5461 is larger than 1000, which is the default sample max length in TimesNet. I will make a PR to make the max length adaptive.