huggingface/transformers

Pull Requests Commits

MekkCyber committed 24 days ago

92e22a65

MekkCyber committed 24 days ago

c75bbf3a

MekkCyber committed 24 days ago

6a913600

Merge remote-tracking branch 'upstream/main' into rm_last_kernel

MekkCyber committed 24 days ago

013d8991

Revert "🚨🚨🚨🚨🚨🚨🚨🚨🚨 default to `"auto"` dtype (#34919)"

LysandreJik committed 24 days ago

a8f32a0e

[`Padding-Free Attention`] Fix packed FA attention with pos ids only (#42801)

vasqu committed 24 days ago

Verified d1eda63f

🚨🚨🚨🚨🚨🚨🚨🚨🚨 default to `"auto"` dtype (#34919)

ArthurZucker committed 24 days ago

Verified f5aa90d0

MekkCyber committed 25 days ago

daf34c55

MekkCyber committed 25 days ago

7164d652

rm the kernel folder !

MekkCyber committed 25 days ago

aec96745

MekkCyber committed 25 days ago

4821e410

feat(granitemoe*): Remove logits upcast when computing loss (#42753)

gabe-l-hart committed 25 days ago

Verified 0af2381f

allow registration of custom checkpoint conversion mappings (#42634)

winglian committed 25 days ago

Verified 45d8168e

Merge remote-tracking branch 'upstream/HEAD' into for-a-green-ci

MekkCyber committed 25 days ago

7a656b8d

MekkCyber committed 25 days ago

8d2f6483

Fully deprecate AutoGPTQ and AutoAWQ for GPT-QModel (#41567)

Qubitium committed 25 days ago

Verified 8ebfd84f

[core] fix fp-quant (#42613)

MekkCyber committed 25 days ago

Verified c3acdd57

[Quantization] Fix Static FP8 Quantization (#42775)

MekkCyber committed 25 days ago

Verified 15735a43

Override Transformers defaults by GGUF defaults (#42770)

a4lg committed 25 days ago

Verified 51a66739

fix: support tensor labels in DataCollatorWithFlattening (#42620)

hqkqn32 committed 25 days ago

Verified f54647c8

Command-a-vision fix (#42642)

dongluw committed 25 days ago

Verified 1b8ccf1c

Stricter checks for mistral patch (#42743)

ArthurZucker committed 25 days ago

Verified b9951b4e

[kernels] Fix kernel CI (#42764)

MekkCyber committed 25 days ago

Verified 508a9764

Ensure e_score_correction_bias dtype of DeepSeek-V3/R1 is FP32 (#42580)

xin3he committed 25 days ago

Verified 2e29a9a6

[Quantization] Fixing some tests (#42763)

MekkCyber committed 25 days ago

Verified 8f978e5b

Only call `torch.autocast` if it will have an effect (#42747)

hmellor committed 26 days ago

Verified 6d0adb5b

Only default `rope_parameters` to empty `dict` if there is something to put in it (#42651)

hmellor committed 26 days ago

Verified 3230fb50

fix links in `CONTRIBUTING.md` (#42745)

casinca committed 26 days ago

Verified 471d7ce9

Add an alternative scenario to EoMT `post_process_semantic_segmentation` in case `path_offsets` is None (#42716)

simonreise committed 26 days ago

Verified 5b4d72c5

🚨🚨 [saving] Default to 50GB shards, and remove non-safe serialization (#42734)

Cyrilvallez committed 26 days ago

Verified 3f3cae74

Older