GitLab

[Refactor] Alias fix

enables tasks of different group but same task_alis (for example, if evaluating on different versions of MMLU

[Refactor] fixes for alternative MMLU tasks.

[Refactor] Logging fixes

[Refactor] Remove deprecated `gold_alias` task YAML option

[Refactor] Mmlu subgroups and weight avg

[Refactor] Bugfix: AttributeError: 'Namespace' object has no attribute 'verbose'

[Refactor] Model written eval

[Refactor] Don't load MMLU auxiliary_train set

[Refactor] Update README, documentation

[Refactor] Describe local dataset usage in docs

[Refactor]fix two bugs when ran with qasper_bool and toxigen