huggingface/transformers

Pull Requests Commits

qgallouedec committed 15 hours ago

f3e0be81

Merge branch 'fix-ep-training-gradients' into fix-ep-through-trainer

qgallouedec committed 15 hours ago

38dd0f6e

Fix expert parallelism through Trainer

qgallouedec committed 15 hours ago

7e2a52a2

qgallouedec committed 17 hours ago

e79ef7a6

Fix wrong gradients for all non-expert parameters in expert-parallel training

qgallouedec committed 17 hours ago

4a0dc8fb

Fix NaN gradients in expert-parallel training: mask uninitialized grouped_mm rows

qgallouedec committed 17 hours ago

7460d407

[TableTransformer, PI0] Fix stale expected values and OOM in integration tests (#48198)

ydshieh committed 21 hours ago

Verified d56c55bf

Post two CI badges on a PR: CPU PR CI and GPU run-slow (#48190)

tarekziade committed 1 day ago

Verified e453228e

Fix stale expected values in integration tests (cuda sm_86 / Aug04 regressions) (#48171)

ydshieh committed 1 day ago

Verified 3ad85ca4

Assign a reviewer even when a codeowner has left, and route models by modality (#48085)

tarekziade committed 1 day ago

Verified e277d1ed

[VITS] Un-skip test_model_forward (#46375)

blipbyte committed 1 day ago

Verified bb593bfb

Apply context parallelism to the evaluation path (#48167)

qgallouedec committed 1 day ago

Verified 695bd682

Fix `gpt_oss` runs on GPU (#48118)

tarekziade committed 1 day ago

Verified 275b5245

🚨 TP dtensor API inference + training (#47579)

3outeille committed 1 day ago

Verified 861f4c41

[EsmFold2] Fix stale expected distogram logit values (#48182)

ydshieh committed 1 day ago

Verified d4c297d1

Fix Apr 05 integration test regressions (cuda sm_86) (#48170)

ydshieh committed 1 day ago

Verified cc33c370

Add MLU support to is_flash_linear_attention_available (#46995)

atri2549 committed 1 day ago

Verified 6a9e2d37

Fix ROCm SDPA-flash skip guard that crashes on RDNA GPUs (#47965)

Abdennacer-Badaoui committed 1 day ago

Verified 3cb4f80f

Fix integration test expected values for cuda sm_86 (Mar 15 regressions) (#48168)

ydshieh committed 1 day ago

Verified 251c2d4f

Fix DynamicCache reconstruction during ExecuTorch export (#47900)

eladsegal committed 1 day ago

Verified 3cd1a230

Revert "Support per-layer cache configuration and attention-mask selection" (#48175)

Cyrilvallez committed 1 day ago

Verified 6b3a9de1

Support per-layer cache configuration and attention-mask selection (#47901)

eladsegal committed 1 day ago

Verified c84ac341

[LLaVA] Fix pixtral integration tests for cuda sm_86 (#48166)

ydshieh committed 1 day ago

Verified 945dac91

Port ESMC and ESMFold2 to Transformers (#46419)

Rocketknight1 committed 1 day ago

Verified c8df3fc9

[Qwen2.5-Omni] Update stale expected values for cuda sm_86 (#48164)

ydshieh committed 1 day ago

Verified cdea840f

[Mistral3] Fix batched integration tests: padding_side=left + update expected values (#48161)

ydshieh committed 1 day ago

Verified dd6c64b7

Fix DeepSeek V2 default vocab size (#48159)

hmellor committed 1 day ago

Verified e4dd56a3

[InternVL] Fix stale expected values for Llama integration tests (cuda sm_80) (#48153)

ydshieh committed 2 days ago

Verified 0c928118

Retry transient network errors (RemoteDisconnected) in github_utils (#48124)

ydshieh committed 2 days ago

Verified 95119511

Revert "[Quantization]: Refactor is_quantization_compressed for format-based detection" (#48072)

subin9 committed 2 days ago

Verified 238a2dbd

Older