huggingface/transformers

Pull Requests Commits

tests `tokenizers==0.15.1rc1`

ArthurZucker committed 1 year ago

0e53a95d

Fix auxiliary loss related code in transformers (#28406)

SangbumChoi committed 1 year ago

Verified 3f69f415

RWKV: raise informative exception when attempting to manipulate `past_key_values` (#28600)

gante committed 1 year ago

Verified 948ffff4

Fix `_speculative_sampling` implementation (#28508)

ofirzaf committed 1 year ago

Verified 9efec114

Allow add_tokens for ESM (#28535)

Rocketknight1 committed 1 year ago

Verified d1578159

[`Llava`] Fix convert_llava_weights_to_hf.py script (#28570)

isaac-vidas committed 1 year ago

Verified 5b7f4bc6

[SigLIP] Don't pad by default (#28578)

NielsRogge committed 1 year ago

Verified faf03541

Fix wrong xpu device in DistributedType.MULTI_XPU mode (#28386)

faaany committed 1 year ago

Verified 8db64367

[Whisper] Finalize batched SOTA long-form generation (#27658)

patrickvonplaten committed 1 year ago

Verified 690fe73f

feat: Sequential beam search (#26304)

Saibo-creator committed 1 year ago

Verified d4fc1eb4

Add w2v2bert to pipeline (#28585)

ylacombe committed 1 year ago

Verified 268fc1fd

amyeroberts committed 1 year ago

b2748a6e

Don't save `processor_config.json` if a processor has no extra attribute (#28584)

ydshieh committed 1 year ago

Verified db9a7e9d

Making CTC training example more general (#28582)

ylacombe committed 1 year ago

Verified 772307be

[Whisper] Fix audio classification with weighted layer sum (#28563)

sanchit-gandhi committed 1 year ago

Verified 186aa6be

[Whisper Tok] Move token ids to CPU when computing offsets (#28485)

sanchit-gandhi committed 1 year ago

Verified 619ecfe2

[ASR Pipe] Update init to set model type and subsequently call parent init method (#28486)

sanchit-gandhi committed 1 year ago

Verified 0eaa5ea3

Fix the documentation checkpoint for xlm-roberta-xl (#28567)

jeremyfowers committed 1 year ago

Verified c662c78c

Use `LoggingLevel` context manager in 3 tests (#28575)

ydshieh committed 1 year ago

Verified 0754217c

Add new meta w2v2-conformer BERT-like model (#28165)

ylacombe committed 1 year ago

Verified d2cdefb9

chore: Fix multiple typos (#28574)

hugo-syn committed 1 year ago

Verified 5d8eb93e

[`Core Tokenization`] Support a fix for spm fast models (#26678)

ArthurZucker committed 1 year ago

Verified 81899778

Use `weights_only` only if torch >= 1.13 (#28506)

ydshieh committed 1 year ago

Verified a1668cc7

Save `Processor` (#27761)

ydshieh committed 1 year ago

Verified 3005f965

Fix Switch Transformers When sparse_step = 1 (#28564)

agemagician committed 1 year ago

Verified 98dda8ed

Allow to train dinov2 with different dtypes like bf16 (#28504)

StarCycle committed 1 year ago

Verified fa6d12f7

Fix SDPA tests (#28552)

fxmarty committed 1 year ago

Verified 2c1eebc1

Add qwen2 (#28436)

JustinLin610 committed 1 year ago

Verified d6ffe74d

Fixes default value of `softmax_scale` in `PhiFlashAttention2`. (#28537)

gugarosa committed 1 year ago

Verified d93ef7d7

symbolic_trace: add past_key_values, llama, sdpa support (#28447)

fxmarty committed 1 year ago

Verified a6adc05e

Older