* fixed formatting based on pre-commit tests
* fixed typo in comments; reduced the number of models in the eval config
* fixed a bug in LLMCompressionDataset, where setting samples=None would result in passing test[:None] to load_dataset
* removed unnecessary variable in _format_table_pivot; changed lark_reporter message to English
* TabMWP
* TabMWP
* fixed
* fixed
* fixed
* done
* done
* done
* add new subjective judgement
* add new subjective judgement
* add new subjective judgement
* add new subjective judgement
* add new subjective judgement
* modified to a more general way
* modified to a more general way
* final
* final
* add summarizer
* add new summarize
* fixed
* fixed
* fixed
---------
Co-authored-by: caomaosong <caomaosong@pjlab.org.cn>