GitLab

[Refactor] Handle `cuda:0` device assignment

Add PEFT, quantization, remote code, LLaMA fix

[Refactor] Fix sample logging

fixed super_glue and removed unused yaml config

[Refactor] Add caveats to `parallelize=True` docs

[Refactor] HellaSwag YAML fix

[Refactor] Log request arguments to per-sample json

[Refactor] Misc. cleanup of dead code

[Refactor] `device_map` options for `hf` model type

[Refactor] Update task statuses on tracking list

[Refactor] batch generation better for `hf` model ; deprecate `hf-causal` in new release

[Refactor] CachingLM support via `--use_cache`