NielsRogge/transformers

Pull Requests Commits

Add changes for uroman package to handle non-Roman characters (#32404)

nandwalritik committed 1 year ago

Verified a378a54a

mps: add `isin_mps_friendly`, a wrapper function for `torch.isin` (#33099)

gante committed 1 year ago

Verified 72d4a3f9

Test: add higher `atol` in `test_forward_with_num_logits_to_keep` (#33093)

gante committed 1 year ago

Verified 894d421e

CI: add torchvision to the consistency image (#32941)

gante committed 1 year ago

Verified 93e0e1a8

support qwen2-vl (#32318)

simonJJJ committed 1 year ago

Verified 19e6e80e

Updated the custom_models.md changed cross_entropy code (#33118)

S-M-J-I committed 1 year ago

Verified 8defc95d

Update Jinja docs with new functions and general cleanup (#33097)

Rocketknight1 committed 1 year ago

Verified 0a7af19f

added doctring to SchedulerType class (#32898)

Arunprakash-A committed 1 year ago

Verified e3a5f35c

DeviceGuard added to use Deformable Attention more safely on multi-GPU (#32910)

DonggeunYu committed 1 year ago

Verified 1dbd9d36

Enable some Jinja extensions and add datetime capabilities (#32684)

Rocketknight1 committed 1 year ago

Verified 371b9c14

Integrate Liger (Linkedin GPU Efficient Runtime) Kernel to Trainer (#32860)

JasonZhu1313 committed 1 year ago

Verified adb91179

Forbid `PretrainedConfig` from saving `generate` parameters; Update deprecations in `generate`-related code 🧹 (#32659)

gante committed 1 year ago

Verified 970a16ec

Reducing memory usage: removing useless logits computation in generate() (#31292)

Cyrilvallez committed 1 year ago

Verified 22e6f145

docs: fix outdated link to TF32 explanation (#32947)

anakin87 committed 1 year ago

Verified d806fa3e

Generate: Deprecate returning legacy cache by default; Handle `use_cache=False` (#32863)

gante committed 1 year ago

Verified a26de151

🌐 [i18n-KO] Translated `knowledge_distillation_for_image_classification.md to Korean" (#32334)

JinukHong committed 1 year ago

Verified 09e6579d

Fix regression on `Processor.save_pretrained` caused by #31691 (#32921)

leloykun committed 1 year ago

Verified 273c0afc

[run_slow] idefics2 (#32840)

andimarafioti committed 1 year ago

Verified 18199b34

Gemma2: eager attention by default (#32865)

gante committed 1 year ago

Verified 975b988b

fix: (issue #32689) `AttributeError` raised when using `Trainer` with `eval_on_start=True` in Jupyter Notebook. (#32849)

fshp971 committed 1 year ago

Verified f1d822ba

Add chat_template for tokenizer extracted from GGUF model (#32908)

Isotr0py committed 1 year ago

Verified ee8c01f8

Improve greedy search memory usage (#32895)

regisss committed 1 year ago

Verified 99d67f1a

Fix benchmark script (#32635)

ydshieh committed 1 year ago

Verified bf97d4aa

Add SynCode to llm_tutorial (#32884)

shubhamugare committed 1 year ago

Verified 92824136

FIX / Hub: Also catch for `exceptions.ConnectionError` (#31469)

younesbelkada committed 1 year ago

Verified eeea7120

CI: separate step to download nltk files (#32935)

gante committed 1 year ago

Verified 8b94d28f

FEAT / Trainer: Add adamw 4bit optimizer (#31865)

SunMarc committed 1 year ago

Verified c42d2645

fix: no need to dtype A in jamba (#32924)

xgal committed 1 year ago

Verified 6baa6f27

fix: Added missing `huggingface_hub` installation to workflows (#32891)

Sai-Suraj-27 committed 1 year ago

Verified af638c4a

Jamba: update integration tests (#32250)

gante committed 1 year ago

Verified f6e2586a

Newer Older