huggingface/transformers

Pull Requests Commits

fix: use mtime by default in Trainer._rotate_checkpoints with automatic fallback (#37260)

Jerry-Terrasse committed 1 year ago

Verified dac44341

Add GGUF support to Gemma3 Text backbone (#37424)

Isotr0py committed 1 year ago

Verified 6daec12d

Llama Kernel integration (#37092)

MekkCyber committed 1 year ago

Verified 0ea11512

Fix require_read_token (#37422)

MekkCyber committed 1 year ago

Verified 9c0c323e

Correctly drop tokens in SwitchTransformer (#37123)

mario-aws committed 1 year ago

Verified bde41d69

Add image classifier donut & update loss calculation for all swins (#37224)

eljandoubi committed 1 year ago

Verified 7ecc5b88

Quark Quantization gated repo (#37412)

MekkCyber committed 1 year ago

Verified 5ae9b2ca

Fix new failure reports not including anything other than `tests/models/` (#37415)

ydshieh committed 1 year ago

Verified d9e76656

[chat-template] Unify tests and clean up 🧼 (#37275)

zucchini-nlp committed 1 year ago

Verified 1ae8d54b

use `rms_norm_eps` for the L2Norm for Llama4 (#37418)

ArthurZucker committed 1 year ago

Verified 10144ff1

Allow rocm systems to run these tests (#37278)

ivarflakstad committed 1 year ago

Verified aa478567

from_pretrained should handle xpu case (#37382)

sywangyi committed 1 year ago

Verified ae5ce226

Send trainer/fsdp/deepspeed CI job reports to a single channel (#37411)

ydshieh committed 1 year ago

Verified 4f139f5a

update `kernels` to 0.4.3 (#37419)

ArthurZucker committed 1 year ago

Verified a2c2fb01

mark llama4 as not supported with fa2 (#37416)

winglian committed 1 year ago

Verified 0ddad2d6

Offloaded hybrid cache for Llama4 (#37401)

Cyrilvallez committed 1 year ago

Verified fbb2054e

Fix Llama4 offset (#37414)

Cyrilvallez committed 1 year ago

Verified 6d8b0b33

Restrict & Explain tp_plan for FBgemm (#37404)

MekkCyber committed 1 year ago

Verified f5865d32

Handle torch ver in flexattn (#37400)

Kh4L committed 1 year ago

Verified e39c7326

Add warning when failed to acquire other user's lock at model download (#37395)

manueldeprada committed 1 year ago

Verified bc0150bb

handle torch version edge cases (#37399)

winglian committed 1 year ago

Verified 9cda4265

the fix that did not get in (#37370)

ArthurZucker committed 1 year ago

Verified e032d12e

Attention Quantization with FBGemm & TP (#37384)

MekkCyber committed 1 year ago

Verified f834ca2c

Fix some failing AWQ tests (#37383)

DerekLiu35 committed 1 year ago

Verified c5c648dd

Apply torchfix to replace deprecated functions: `_pytree._register_pytree_node` and `torch.cpu.amp.autocast` (#37372)

bzhong-solink committed 1 year ago

Verified 71b35387

Fix warning message for PEFT models in text-generation pipeline #36783 (#36887)

falconlee236 committed 1 year ago

Verified ad340908

Add "selecting a quantization method" doc (#37159)

DerekLiu35 committed 1 year ago

Verified 2527f71a

update deepspeed docker (#37371)

SunMarc committed 1 year ago

Verified 7ae0be72

Add glm4 (#37388)

ArthurZucker committed 1 year ago

Verified e3eda6d1

fix: llama4 conversion script no_rope_layers (#37359)

jmkuebler committed 1 year ago

Verified 1e6ff5fd

Newer Older