huggingface/transformers

Pull Requests Commits

ArthurZucker committed 2 years ago

b1292bca

fix bug:divide by zero in _maybe_log_save_evaluate() (#28251)

frankenliu committed 2 years ago

Verified 87ae2a46

Fix trainer saving safetensors: metadata is None (#28219)

hiyouga committed 2 years ago

Verified 502a10a6

Update docs around mixing hf scheduler with deepspeed optimizer (#28223)

dwyatte committed 2 years ago

Verified cad9f5c6

small typo (#28229)

stas00 committed 2 years ago

Verified 3cefac1d

fix FA2 when using quantization (#28203)

pacman100 committed 2 years ago

Verified 3b7675b2

[`Awq`] Enable the possibility to skip quantization for some target modules (#27950)

younesbelkada committed 2 years ago

Verified fa21ead7

[`Llava`] Fix llava index errors (#28032)

younesbelkada committed 2 years ago

Verified 29e7a1e1

update the logger message with accordant weights_file_name (#28181)

izyForever committed 2 years ago

Verified 68fa1e85

Fixing visualization code for object detection to support both types of bounding box. (#27842)

Anindyadeep committed 2 years ago

Verified 74d9d0ce

[Whisper] Fix word-level timestamps with bs>1 or num_beams>1 (#28114)

ylacombe committed 2 years ago

Verified 5da3db3f

Drop `feature_extractor_type` when loading an image processor file (#28195)

ydshieh committed 2 years ago

Verified c4df7c16

Fix the check of models supporting FA/SDPA not run (#28202)

ydshieh committed 2 years ago

Verified bb3bd447

Bug: `training_args.py` fix missing import with accelerate with version `accelerate==0.20.1` (#28171)

michaelfeil committed 2 years ago

Verified e37ab52d

Add Swinv2 backbone (#27742)

NielsRogge committed 2 years ago

Verified c9fb250a

Fix: [SeamlessM4T - S2TT] Bug in batch loading of audio in torch.Tensor format in the SeamlessM4TFeatureExtractor class (#27914)

nicholasneo78 committed 2 years ago

Verified 1ef86c4f

Fix ONNX export for causal LM sequence classifiers by removing reverse indexing (#28144)

dwyatte committed 2 years ago

Verified 548a8f61

Update `docs/source/en/perf_infer_gpu_one.md` (#28198)

ydshieh committed 2 years ago

Verified 71f46057

[`Docs`] Add 4-bit serialization docs (#28182)

younesbelkada committed 2 years ago

Verified 3a8769f6

Update YOLOS slow test values (#28187)

amyeroberts committed 2 years ago

Verified 3657748b

Fix slow backbone tests - out_indices must match stage name ordering (#28186)

amyeroberts committed 2 years ago

Verified cd1350ce

Even more TF test fixes (#28146)

Rocketknight1 committed 2 years ago

Verified 260b9d21

[`Mixtral` & `Mistral`] Add support for sdpa (#28133)

ArthurZucker committed 2 years ago

Verified f9a98c47

[Whisper] Use torch for stft if available (#26119)

sanchit-gandhi committed 2 years ago

Verified 814619f5

Fix `input_embeds` docstring in encoder-decoder architectures (#28168)

gante committed 2 years ago

Verified 7e93ce40

[bnb] Let's make serialization of 4bit models possible (#26037)

poedator committed 2 years ago

Verified 4f7806ef

disable test_retain_grad_hidden_states_attentions on SeamlessM4TModelWithTextInputTest (#28169)

dwyatte committed 2 years ago

Verified e268d7e5

Fix yolos resizing (#27663)

amyeroberts committed 2 years ago

Verified 1d777359

Generate: fix speculative decoding (#28166)

gante committed 2 years ago

Verified 45b70384

[docs] Trainer docs (#28145)

stevhliu committed 2 years ago

Verified 01c081d1

Older