Releases · NovaSky-AI/SkyThought

What's Changed

docs: update data/README.md by @eltociear in #3
Update numina by @DachengLi1 in #18
add bon inference and check by @caoshiyi in #19
Fix AIME & MATH500 prompt format by @lynnliu030 in #25
[Evals] Add eval support for GSM8K, ARC-C and MMLU-PRO by @SumanthRH in #21
[Doc] Add Chat Demo & Discord Link by @caoshiyi in #36
Upload model by @DachengLi1 in #29
Fix evaluation on MMLUPro, GSM8K, ARC-C. by @fanqiwan in #39
[Evals] Add AMC 23 Eval by @erictang000 in #40
Sky-T1-32B-Flash, Reduce Overthinking by @tyler-griggs in #43
[Evals] Add OlympiadBenchMath (en) by @SumanthRH in #44
[Evals] Use a lower writer batch size for livecodebench by @SumanthRH in #45
[Evals] Added Minevera Math by @kouroshHakha in #41
[Evals] Add results on non-reasoning tasks by @SumanthRH in #26
docs(train/README): intended links by @guspan-tanadi in #8
[Evals] Refactor tasks, add linting and CI by @SumanthRH in #47
[Evals] Update base_instruct_evals.md reference by @emmanuel-ferdman in #61
Update README.md by @tyler-griggs in #62
Minor fixes after #23 by @SumanthRH in #58
Refactor model-specific configs and move data curation scripts by @SumanthRH in #60
[evals] Add support for scaling evals and inference with ray by @erictang000 in #63
chore(base_instruct_evals): access branch links by @guspan-tanadi in #55
Navigate installed package relative links by @guspan-tanadi in #68
Fix reproducibility issues, save metrics to disk and cleanup scripts by @SumanthRH in #67
Add Sky-T1-7B by @caoshiyi in #72
Minor fixes by @SumanthRH in #75
New package CLI by @SumanthRH in #71
Fixes before release by @SumanthRH in #77
[Bugfix] Rename module in tests by @SumanthRH in #78
Add YAMLs for numina subsets and fix broken links by @SumanthRH in #79