HuggingFace Datasets Integration

This release integrates HuggingFace datasets as the core dataset management interface, removing previous custom downloaders.

What's Changed

• Refactor Task downloading to use HuggingFace.datasets by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/300
• Add templates and update docs by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/308
• Add dataset features to TriviaQA by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/305
• Add SWAG by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/306
• Fixes for using lm_eval as a library by @dirkgr in https://github.com/EleutherAI/lm-evaluation-harness/pull/309
• Researcher2 by @researcher2 in https://github.com/EleutherAI/lm-evaluation-harness/pull/261
• Suggested updates for the task guide by @StephenHogg in https://github.com/EleutherAI/lm-evaluation-harness/pull/301
• Add pre-commit by @Mistobaan in https://github.com/EleutherAI/lm-evaluation-harness/pull/317
• Decontam import fix by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/321
• Add bootstrap_iters kwarg by @Muennighoff in https://github.com/EleutherAI/lm-evaluation-harness/pull/322
• Update decontamination.md by @researcher2 in https://github.com/EleutherAI/lm-evaluation-harness/pull/331
• Fix key access in squad evaluation metrics by @konstantinschulz in https://github.com/EleutherAI/lm-evaluation-harness/pull/333
• Fix make_disjoint_window for tail case by @richhankins in https://github.com/EleutherAI/lm-evaluation-harness/pull/336
• Manually concat tokenizer revision with subfolder by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/343
• [deps] Use minimum versioning for numexpr by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/352
• Remove custom datasets that are in HF by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/330
• Add TextSynth API by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/299
• Add the original LAMBADA dataset by @jon-tow in https://github.com/EleutherAI/lm-evaluation-harness/pull/357

New Contributors

• @dirkgr made their first contribution in https://github.com/EleutherAI/lm-evaluation-harness/pull/309
• @Mistobaan made their first contribution in https://github.com/EleutherAI/lm-evaluation-harness/pull/317
• @konstantinschulz made their first contribution in https://github.com/EleutherAI/lm-evaluation-harness/pull/333
• @richhankins made their first contribution in https://github.com/EleutherAI/lm-evaluation-harness/pull/336

Full Changelog: https://github.com/EleutherAI/lm-evaluation-harness/compare/v0.2.0...v0.3.0