huggingface/transformers

Pull Requests Commits

MekkCyber committed 1 year ago

475bb3ca

MekkCyber committed 1 year ago

981d34c9

AMD DeepSpeed image additional HIP dependencies (#36195)

ivarflakstad committed 1 year ago

Verified 936aeb70

Fix `LlavaForConditionalGenerationModelTest::test_config` after #36077 (#36230)

ydshieh committed 1 year ago

Verified 23d6095e

[tests] fix `EsmModelIntegrationTest::test_inference_bitsandbytes` (#36225)

faaany committed 1 year ago

Verified fae0f3dd

set `test_torchscript = False` for Blip2 testing (#35972)

ydshieh committed 1 year ago

Verified dd16acb8

Use `args.num_workers` in `check_modular_conversion.py` (#36200)

ydshieh committed 1 year ago

Verified 0a9923a6

add shared experts for upcoming Granite 4.0 language models (#35894)

mayank31398 committed 1 year ago

Verified a570e2ba

Add @require_bitsandbytes to Aria test_batched_generation (#36192)

ivarflakstad committed 1 year ago

Verified 7ae7e87a

[Bugfix] Fix reloading of pixtral/llava configs (#36077)

kylesayrs committed 1 year ago

Verified bcfc9d79

🔴 VLM: compile compatibility (#35724)

zucchini-nlp committed 1 year ago

Verified 0c78ef6c

Guard against unset resolved_archive_file (#35628)

dmlap committed 1 year ago

Verified b45cf0e9

Revert qwen2 breaking changes related to attention refactor (#36162)

ArthurZucker committed 1 year ago

Verified 96f01a36

Add require_read_token to fp8 tests (#36189)

MekkCyber committed 1 year ago

Verified cb586a39

New HIGGS quantization interfaces, JIT kernel compilation support. (#36148)

BlackSamorez committed 1 year ago

Verified 5f726f8b

Prepare processors for VideoLLMs (#36149)

zucchini-nlp committed 1 year ago

Verified 15ec971b

Add ImageProcessorFast to Qwen2.5-VL processor (#36164)

Isotr0py committed 1 year ago

Verified 33d1d715

Chat template docs (#36163)

zucchini-nlp committed 1 year ago

Verified 1931a351

CI: fix `test-save-trainer` (#36191)

zucchini-nlp committed 1 year ago

Verified 3bf02cf4

Add support for partial rotary embeddings in Phi3 model (#35947)

garg-amit committed 1 year ago

Verified 0ae93d31

Uniformize OwlViT and Owlv2 processors (#35700)

yonigozlan committed 1 year ago

Verified 336dc69d

Fix make_batched_videos and add tests (#36143)

yonigozlan committed 1 year ago

Verified e6a79817

Fix a mistake in #36175 (#36179)

ydshieh committed 1 year ago

Verified 8fd4bc7d

Follow up to SpQR integration (#36176)

MekkCyber committed 1 year ago

Verified b1a2de07

Fix the key name for _load_rng_state under torch.cuda (#36138)

wizyoung committed 1 year ago

Verified 12962fe8

Make `check_repository_consistency` run faster by MP (#36175)

ydshieh committed 1 year ago

Verified bfe46c98

Optimize Qwen2VL vision model by precomputing cos/sin embeds before ViT blocks (#35837)

li-plus committed 1 year ago

Verified 5f0fd118

Use tqdm auto (#35726)

SmartManoj committed 1 year ago

Verified d72642bc

CI: avoid human error, automatically infer generative models (#33212)

gante committed 1 year ago

Verified 62c7ea02

add disable compile option (#36161)

ArthurZucker committed 1 year ago

Verified 06231fdf

Older