huggingface/transformers

Pull Requests Commits

fix rope modeling to follow qwen2 vl instead + change auto loading to specifically load via pretrained (overridable from pretrained for auto classes)

vasqu committed 9 days ago

14b2bf3f

vasqu committed 9 days ago

7aba4635

Merge branch 'main' into ernie_vl

vasqu committed 9 days ago

2468a6f1

Sam: Perception Encoder Audiovisual (#42905)

eustlb committed 9 days ago

Verified 9aef5ca4

add small dummy integration tests

vasqu committed 9 days ago

25757d5e

Merge branch 'main' into ernie_vl

vasqu committed 9 days ago

Verified 953520f0

fix `Dtensor` and `tensor` mismatch (#42906)

3outeille committed 9 days ago

Verified b1a2fba1

processor path change

vasqu committed 9 days ago

ed99cd06

vasqu committed 9 days ago

eeb68ddd

refactor more tokenizers - v5 guide update (#42768)

itazap committed 9 days ago

Verified 6994c5ac

fix video processing loading logic

vasqu committed 9 days ago

3e215d51

fix: typehits for Causal LM models (#42885)

CandiedCode committed 9 days ago

Verified 4c64a8fb

Remove duplicated processor class from config (#42806)

zucchini-nlp committed 9 days ago

Verified b61da251

Add local kernel loading support to KernelConfig(). (#42800)

zheliuyu committed 9 days ago

Verified 24275124

Simplify tie weights logic (#42895)

Cyrilvallez committed 9 days ago

Verified 4d6516e2

fix FastSpeech2ConformerTokenizer crash in tokenize (#42888)

sywangyi committed 9 days ago

Verified 24b311ee

Added kernels from kernel hub for Bamba model (#41540)

romitjain committed 9 days ago

Verified 0f896619

Merge main into ernie_vl

vasqu committed 9 days ago

cb1a33ca

revert a few tok things (additional_special_tokens), updated conversion

vasqu committed 9 days ago

f8f46465

Fix GraniteMoeHybrid in transformers v5 (#42872)

avihu111 committed 9 days ago

Verified 5d2f82b5

[`Ernie 4.5 Moe`] Fix routing, weights, and update expectations (#42653)

vasqu committed 9 days ago

Verified 4e7cecb2

Fix convert_tekken_tokenizer (#42592)

juliendenize committed 9 days ago

Verified 252afd89

Stop collecting all model parameters to save models when using DeepSpeed and LoRA (#41416)

Taise228 committed 9 days ago

Verified 89998bdd

[modular] Fix a weird renaming edge-case (#42844)

Cyrilvallez committed 9 days ago

Verified 8d526c23

[Devstral] Make sure FP8 conversion works correctly (#42715)

patrickvonplaten committed 9 days ago

Verified 7960b5ea

Add missing ModelOutput subclass return type hints (#41219)

tomaarsen committed 9 days ago

Verified 6c7c992f

Fix BLT training_ci overfit test (#42685)

preetam1407 committed 9 days ago

Verified 0f97c688

Add `.on_push_begin()` callback to Trainer and implement for `TrackioCallback` (#42850)

abidlabs committed 9 days ago

Verified 7f52a2a4

[docs] optimizations quickstart (#42538)

stevhliu committed 10 days ago

Verified 31de95ef

vasqu committed 10 days ago

2455aab8

Older