NielsRogge/transformers

Pull Requests Commits

Add w2v2bert to pipeline (#28585)

ylacombe committed 2 years ago

Verified 268fc1fd

amyeroberts committed 2 years ago

b2748a6e

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

ydshieh committed 2 years ago

Verified db9a7e9d

Making CTC training example more general (#28582)

ylacombe committed 2 years ago

Verified 772307be

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

sanchit-gandhi committed 2 years ago

Verified 186aa6be

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

sanchit-gandhi committed 2 years ago

Verified 619ecfe2

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

sanchit-gandhi committed 2 years ago

Verified 0eaa5ea3

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

jeremyfowers committed 2 years ago

Verified c662c78c

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

ydshieh committed 2 years ago

Verified 0754217c

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

ylacombe committed 2 years ago

Verified d2cdefb9

chore: Fix multiple typos (#28574)

hugo-syn committed 2 years ago

Verified 5d8eb93e

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

ArthurZucker committed 2 years ago

Verified 81899778

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

ydshieh committed 2 years ago

Verified a1668cc7

Save `Processor` (#27761)

ydshieh committed 2 years ago

Verified 3005f965

Fix Switch Transformers When sparse_step = 1 (#28564)

agemagician committed 2 years ago

Verified 98dda8ed

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

StarCycle committed 2 years ago

Verified fa6d12f7

Fix SDPA tests (#28552)

fxmarty committed 2 years ago

Verified 2c1eebc1

Add qwen2 (#28436)

JustinLin610 committed 2 years ago

Verified d6ffe74d

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

gugarosa committed 2 years ago

Verified d93ef7d7

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

fxmarty committed 2 years ago

Verified a6adc05e

[Makefile] Exclude research projects from format (#28551)

patrickvonplaten committed 2 years ago

Verified 09eb11a1

Config: warning when saving generation kwargs in the model config (#28514)

gante committed 2 years ago

Verified f4f57f9d

Add is_model_supported for fx (#28521)

inisis committed 2 years ago

Verified 7142bdfa

Clearer error for SDPA when explicitely requested (#28006)

fxmarty committed 2 years ago

Verified 02f8738e

[`SpeechT5Tokenization`] Add copied from and fix the `convert_tokens_to_string` to match the fast decoding scheme (#28522)

ArthurZucker committed 2 years ago

Verified fe23256b

[`TokenizationRoformerFast`] Fix the save and loading (#28527)

ArthurZucker committed 2 years ago

Verified 96d08831

[ `TokenizationUtils`] Fix `add_special_tokens` when the token is already there (#28520)

ArthurZucker committed 2 years ago

Verified 716df5fb

Fix/speecht5 bug (#28481)

NimaYaqmuri committed 2 years ago

Verified 07ae53e6

Fix mismatching loading in from_pretrained with/without accelerate (#28414)

fxmarty committed 2 years ago

Verified 66db33dd

Improving Training Performance and Scalability Documentation (#28497)

HamzaFB committed 2 years ago

Verified 002566f3

Newer Older