NielsRogge/transformers

Pull Requests Commits

Fix qwen2_5 get_rope_index tensor device locations (#37597)

rphmeier committed 292 days ago

Verified 43bb4c04

updated hidden_features for FlaxDinov2SwiGLUFFN in Dinov2 (#37747)

premmurugan229 committed 292 days ago

Verified dd2649fa

[generate] skip compilation on cpu offload (#37709)

gante committed 292 days ago

Verified 8bdd4f2a

`GPT2Model` StaticCache support (#35761)

poedator committed 292 days ago

Verified 7c62e693

[cache] fix `HybridCache` init when `device` is passed (#37718)

gante committed 292 days ago

Verified 9f927c82

Expand quantized data type support for tensor parallelism (#37719)

amd-xiaoyu12 committed 292 days ago

Verified 4fee3209

Update `MllamaForConditionalGenerationIntegrationTest` (#37750)

ydshieh committed 292 days ago

Verified 0f7940bb

Skip all `AriaForConditionalGenerationIntegrationTest` on `T4` (#37746)

ydshieh committed 292 days ago

Verified 7e6f36cd

[performance_optim] define flash attention mask on NPU device directly (#37698)

FightingZhen committed 292 days ago

Verified 0327d0f7

Correctly raise errors when downloading tokenizer files (#37740)

Cyrilvallez committed 292 days ago

Verified 14e28bd7

Fix `embeds_to_talker` device in Qwen2.5-Omni (#37739)

BakerBunker committed 292 days ago

Verified 0ec04959

fix: learning_rate logged as tensor causing save issue with deepspeed (#37704)

NanoCode012 committed 292 days ago

Verified 72e48440

[VLMs] fix flash-attention tests (#37603)

zucchini-nlp committed 292 days ago

Verified 1cfcbfca

Make sure torch_is_available before using torch.distributed (#37693)

MekkCyber committed 292 days ago

Verified 02baa61f

[tests] fix `test_nemotron_8b_generation_sdpa` (#37665)

faaany committed 292 days ago

Verified 864e9636

Fix torchao doc examples (#37697)

MekkCyber committed 292 days ago

Verified 9b3bf4a2

Fix inference bugs in Qwen2.5 Omni (#37701)

BakerBunker committed 292 days ago

Verified 3ed56bea

Fix Aria tests (#37444)

jiqing-feng committed 292 days ago

Verified b7f7aa78

Add Fast Image Processor for MobileNetV1 (#37111)

dmdaksh committed 293 days ago

Verified b6d65e40

Add Fast Image Processor for PoolFormer (#37182)

rootonchair committed 293 days ago

Verified dea1919b

Add Fast PVT Processor (#37204)

keetrap committed 293 days ago

Verified b491f128

enable 4 test_trainer cases on XPU (#37645)

yao-matrix committed 293 days ago

Verified 19e9079d

Process inputs directly in apply_chat_template in image-text-to-text pipeline (#35616)

yonigozlan committed 293 days ago

Verified 5cd6b640

[tests, `qwen2_5_omni`] fix flaky tests (#37721)

gante committed 293 days ago

Verified 80ea2c05

Qwen 2.5 Omni: apply video defaults (#37660)

pcuenca committed 293 days ago

Verified 63c63313

[internvl] fix chat template (#37656)

zucchini-nlp committed 293 days ago

Verified 1e908736

TransfoXL is deprecated, don't keep it in tested examples! (#37707)

Rocketknight1 committed 293 days ago

Verified 9ec8be56

[CI] add back `sacrebleu` (and document why) (#37700)

gante committed 293 days ago

Verified be9b0e85

Add maintainers for ROCm/Intel XPU/Ascend NPU (#37678)

Rocketknight1 committed 293 days ago

Verified 1d7d7a94

[cleanup] remove `/model_cards` 🧹 🧹 (#37685)

gante committed 293 days ago

Verified cc9a245e

Newer Older