huggingface/transformers

Pull Requests Commits

Patch release: v4.27.8

sgugger committed 2 years ago

Verified 4e9f6fc6

Revert "Error (also in original) model, scaling only q matrix not qk.T dot product (qk.T/sqrt(dim_per_head))" (#22444)

sgugger committed 2 years ago

Verified 4277b3dd

Patch release: v4.27.3

sgugger committed 2 years ago

Verified 5e3b19a8

Enforce `max_memory` for device_map strategies (#22311)

sgugger committed 2 years ago

Verified 62d9baa5

Patch release: v4.27.2

sgugger committed 2 years ago

Verified 68287689

Fix balanced and auto device_map (#22271)

sgugger committed 2 years ago

Verified 1e39734c

Release: v4.27.1

LysandreJik committed 2 years ago

2355e463

Regression pipeline device (#22190)

sgugger committed 2 years ago

659ef0b5

Revert 22152 MaskedImageCompletionOutput changes (#22187)

amyeroberts committed 2 years ago

36ed7508

Release: v4.27.0

sgugger committed 2 years ago

Verified d941f07a

Revert "Enforce same behavior as PyTorch 2.0 for older versions" (#22163)

sgugger committed 2 years ago

Verified c52c5282

[trainer] add `--optim adamw_torch_fused` for pt-2.0+ (#22144)

stas00 committed 2 years ago

Verified 085bf5c1

to_pil - don't rescale if int and in range 0-255 (#22158)

amyeroberts committed 2 years ago

Verified c6318c37

Create MaskedImageCompletionOutput and fix ViT docs (#22152)

alaradirik committed 2 years ago

Verified 3b22bfbc

Fix big model inference for T5 models in float16 (#22095)

sgugger committed 2 years ago

Verified b45192ec

Translation Italian: perf_train_cpu and perf_train_cpu_many (#22151)

nickprock committed 2 years ago

Verified 7f5ad6c3

Update 2 doctest expected values for torch 2.0.0 (#22148)

ydshieh committed 2 years ago

Verified ff887035

Add ConvNeXT V2 (#21679)

alaradirik committed 2 years ago

Verified cdddfbff

Move `is_pipeline_test_to_skip` to specific model test classes (#21999)

ydshieh committed 2 years ago

Verified 6c2ad00c

[🛠️] Fix-whisper-breaking-changes (#21965)

ArthurZucker committed 2 years ago

Verified 2beabd24

docs: New terms and updates to glossary (#21982)

MichaelRipa committed 2 years ago

Verified 101a6cd2

Prepare daily CI for torch 2.0.0 (#22135)

ydshieh committed 2 years ago

Verified ba9e0191

[Safetensors] Add explicit flag to from pretrained (#22083)

patrickvonplaten committed 2 years ago

Verified f780557a

Remove backend check for torch.compile (#22140)

sgugger committed 2 years ago

Verified 3a35937e

[deepspeed docs] Activation Checkpointing (#22099)

stas00 committed 2 years ago

Verified 618697ef

[trainer] fix bug in grad accum with multiple epochs (#22098)

stas00 committed 2 years ago

Verified 5b85add7

Enforce same behavior as PyTorch 2.0 for older versions (#22136)

sgugger committed 2 years ago

Verified 1c801d65

Trainer: let generate pick its inputs (#22108)

gante committed 2 years ago

Verified e16cbe88

[`Whiper`] add `get_input_embeddings` to `WhisperForAudioClassification` (#22133)

younesbelkada committed 2 years ago

Verified d979cf6e

Update configuration_align.py (projected_dim=640) (#22139)

bishmdl76 committed 2 years ago

Verified 98797237

Older