NielsRogge/transformers

Pull Requests Commits

[`InstructBlip`] Add instruct blip int8 test (#24555)

younesbelkada committed 2 years ago

Verified 33b5ef5c

Fix processor __init__ bug if image processor undefined (#24554)

amyeroberts committed 2 years ago

Verified c70c88a2

[`gpt2-int8`] Add gpt2-xl int8 test (#24543)

younesbelkada committed 2 years ago

Verified 903b97d8

Update `EncodecIntegrationTest` (#24553)

ydshieh committed 2 years ago

Verified b0651655

Update PT/TF weight conversion after #24030 (#24547)

ydshieh committed 2 years ago

Verified 6c57ce15

Fix typing annotations for FSDP and DeepSpeed in TrainingArguments (#24549)

mryab committed 2 years ago

Verified c5e29d43

Allow for warn_only selection in enable_full_determinism (#24496)

Frank995 committed 2 years ago

Verified daccde14

Unpin DeepSpeed and require DS >= 0.9.3 (#24541)

ydshieh committed 2 years ago

Verified 11cb6e0f

⚠️ Time to say goodbye to py37 (#24091)

ydshieh committed 2 years ago

Verified e84bf1f7

Add bitsandbytes support for gpt2 models (#24504)

DarioSucic committed 2 years ago

Verified 12240925

Finishing tidying keys to ignore on load (#24535)

sgugger committed 2 years ago

Verified 89b6ee49

Fix Typo (#24530)

MS Kim(tony9402) committed 2 years ago

Verified 04f46a22

Allow backbones not in backbones_supported - Maskformer Mask2Former (#24532)

amyeroberts committed 2 years ago

Verified 462f77cb

Clean load keys (#24505)

sgugger committed 2 years ago

Verified 8e5d1619

[Mask2Former] Remove SwinConfig (#24259)

NielsRogge committed 2 years ago

Verified 53194991

Fix LR scheduler based on bs from auto bs finder (#24521)

muellerzr committed 2 years ago

Verified fb6a6276

Find module name in an OS-agnostic fashion (#24526)

sgugger committed 2 years ago

Verified 38db04ec

Update `huggingface_hub` commit sha (#24527)

ydshieh committed 2 years ago

Verified 7d150d68

set model to training mode before accelerate.prepare (#24520)

sywangyi committed 2 years ago

Verified 4e8929dc

[`T5`] Add T5ForQuestionAnswering and MT5ForQuestionAnswering (#24481)

sjrl committed 2 years ago

Verified 06910f5a

Update hyperparameter_search.py (#24515)

pacman100 committed 2 years ago

Verified bcf02ec7

use accelerate autocast in jit eval path, since mix precision logic is… (#24460)

sywangyi committed 2 years ago

Verified 6fe8d198

🌐 [i18n-KO] Translated `tflite.mdx` to Korean (#24435)

0525hhgus committed 2 years ago

Verified 0863436b

Fix poor past ci (#24485)

ydshieh committed 2 years ago

Verified 4abd3ee4

Fix TypeError: Object of type int64 is not JSON serializable (#24340)

xiaoli committed 2 years ago

Verified 239ace15

Generate: `min_tokens_to_keep` has to be `>= 1` (#24453)

gante committed 2 years ago

Verified ac19871c

Generate: `group_beam_search` requires `diversity_penalty>0.0` (#24456)

gante committed 2 years ago

Verified 5f3efdf7

🚨🚨 Fix group beam search (#24407)

hukuda222 committed 2 years ago

Verified 43479ef9

Fix link in utils (#24501)

SoyGema committed 2 years ago

Verified 68c92981

Compute `dropout_probability` only in training mode (SpeechT5) (#24498)

ydshieh committed 2 years ago

Verified 7b4e3b5b

Newer Older