Add udop #56
ArthurZucker
merged
283 commits into
NielsRogge:add_udop
from
ArthurZucker:add_udop
add custom RMSNorm to `ALL_LAYERNORM_LAYERS` (#26227)
e3a4bd2b
Keep relevant weights in fp32 when `model._keep_in_fp32_modules` is s…
da971b22
Fix FSMT weight sharing (#26292)
26ba56cc
update hf hub dependency to be compatible with the new tokenizers (#2…
b132c170
Porting the torchaudio kaldi fbank implementation to audio_utils (#26…
9a307534
More error message fixup, plus some linebreaks! (#26296)
000e52ae
[QUICK FIX LINK] Update trainer.py (#26293)
587b7b16
Use CircleCI `store_test_results` (#26223)
06ee91ae
Fix doctest CI (#26324)
c3ecf2d9
[doc] fixed indices in obj detection example (#26343)
dcbfd93d
[`core` ] Integrate Flash attention 2 in most used models (#25598)
368a58e6
[TTA Pipeline] Fix MusicGen test (#26348)
914771cb
Add image to image pipeline (#25393)
576cd45a
feat: adding num_proc to load_dataset (#26326)
910faa3e
Fixed unclosed p tags (#26240)
5936c8c5
Update add_new_model.md (#26365)
6accd5ef
Fix MusicGen logging error (#26370)
0ee45906
[docs] removed MaskFormerSwin and TimmBackbone from the table on inde…
546e7679
Update tiny model information and pipeline tests (#26285)
d9e4bc28
Add Russian localization for README (#26208)
033ec57c
🌐 [i18n-KO] Translated `audio_classification.mdx` to Korean (#26200)
5e09af2a
Add Nougat (#25942)
ace74d16
[ViTMatte] Add resources (#26317)
a09130fe
Deleted duplicate sentence (#26394)
a8531f3b
added support for gradient checkpointing in ESM models (#26386)
6ce6a5ad
Fix DeepSpeed issue with Idefics (#26393)
0ac38750
[InternLM] Add support for InternLM (#26302)
6ba63ac3
Add torch `RMSProp` optimizer (#26425)
408b2b3c
Fix padding for IDEFICS (#26396)
abd25310
Update semantic_segmentation.md (#26419)
777f2243
Fixing tokenizer when `transformers` is installed without `tokenizers…
a0be960d
[`FA` / `tests`] Add use_cache tests for FA models (#26415)
153755ee
add bf16 mixed precision support for NPU (#26163)
946bac79
[`PEFT`] Fix PEFT multi adapters support (#26407)
3ca18d6d
[Mistral] Mistral-7B-v0.1 support (#26447)
72958fcd
Fix failing doctest (#26450)
78dd1202
Update `runs-on` in workflow files (#26435)
6ae71ec8
[i18n-DE] Complete first toc chapter (#26311)
ef81759e
🌐 [i18n-KO] Translated `debugging.md` to Korean (#26246)
a0922a53
🌐 [i18n-KO] Translated `perf_train_gpu_many.md` to Korean (#26244)
ab37b801
optimize VRAM for calculating pos_bias in LayoutLM v2, v3 (#26139)
a7e0ed82
Fix `cos_sin` device issue in Falcon model (#26448)
375b4e09
docs: change assert to raise and some small docs (#26232)
ba47efbf
change mention of decoder_input_ids to input_ids and same with decode…
098c3f40
[VITS] Fix speaker_embed device mismatch (#26115)
52e2c13d
[`PEFT`] introducing `adapter_kwargs` for loading adapters from diffe…
38e96324
Do not warn about unexpected decoder weights when loading T5EncoderMo…
216dff75
fix_mbart_tied_weights (#26422)
5e11d72d
Esm checkpointing (#26454)
4e931a8e
[Whisper Tokenizer] Make decoding faster after adding timestamps (#26…
211f93aa
[docs] Update offline mode docs (#26478)
7bb1c0c1
[docs] navigation improvement between text gen pipelines and text gen…
14170b78
Skip 2 failing persimmon pipeline tests for now (#26485)
9b23d0de
Avoid all-zeor attnetion mask used in testing (#26469)
39117744
[Flax Examples] Seq2Seq ASR Fine-Tuning Script (#21764)
68e85fc8
[ASR Pipe] Improve docs and error messages (#26476)
0b192de1
Revert falcon exception (#26472)
67239f73
Fix num_heads in _upad_input (#26490)
ca0379b8
Fix requests connection error during modelcard creation (#26518)
7d77d7f7
Fix issue of canine forward requiring input_ids anyway (#26290)
6d02ca4b
Fix broken link to video classification task (#26487)
7d6627d0
[`PEFT`] Pass token when calling `find_adapter_config` (#26488)
24178c24
[`core`/ `auto` ] Fix bnb test with code revision + bug with code re…
6824461f
Fix model integration ci (#26322)
63864e05
[`PEFT`] Protect `adapter_kwargs` check (#26537)
1b8decb0
Remove-warns (#26483)
e4dad4fe
[Doctest] Add configuration_roformer.py (#26530)
4b4c6aab
Code-llama-nit (#26300)
bab33319
add build_inputs_with_special_tokens to LlamaFast (#26297)
c20d90d5
🌐 [i18n-KO] Translated `tokenizer_summary.md` to Korean (#26243)
1470f731
[i18n-DE] contribute chapter (#26481)
9ed538f2
![dependabot[bot]](https://avatars.githubusercontent.com/in/29110?v=4&s=60)
Bump urllib3 from 1.26.5 to 1.26.17 in /examples/research_projects/lx…
e092b4ad
Bump urllib3 from 1.26.5 to 1.26.17 in /examples/research_projects/vi…
6de6fdd0
Bump urllib3 from 1.26.9 to 1.26.17 in /examples/research_projects/de…
cf345d5f
[RFC, Logging] Change warning to info (#26545)
df6a855e
Add tokenizer kwargs to fill mask pipeline. (#26234)
b5ca8fcd
[Wav2Vec2 and Co] Update init tests for PT 2.1 (#26494)
768aa3d9
[AMD] Add initial version for run_tests_multi_gpu (#26346)
3632fb3c
[Doctest] Add `configuration_encoder_decoder.py` (#26519)
245da7ed
Nit-added-tokens (#26538)
1a2e966c
[`Mistral`] Add Flash Attention-2 support for `mistral` (#26464)
ae9a344c
[`PEFT`] Final fixes (#26559)
2aef9a96
[`Nougat`] from transformers import * (#26562)
c26b2a29
v4.35.0.dev0
bd620591
[Whisper] Allow basic text normalization (#26149)
57f44dc4
🌐 [i18n-KO] Translated `semantic_segmentation.md` to Korean (#26515)
2c7b26f5
[Tokenizers] Skip tests temporarily (#26574)
5c66378c
docs: feat: add clip notebook resources from OSSCA community (#26505)
2f3ea08a
Bump pillow from 9.3.0 to 10.0.1 in /examples/research_projects/decis…
fc296f41
Extend Trainer to enable Ascend NPU to use the fused Adamw optimizer …
4fdf47cd
feat: add trainer label to wandb run upon initialization (#26466)
122b2657
Docstring check (#26052)
03af4c42
Add add_generation_prompt argument to apply_chat_template (#26573)
8b46c5bc
refactor: change default block_size (#26229)
6015f91a
[Mistral] Update config docstring (#26593)
0a49f909
Add # Copied from statements to audio feature extractors that use the…
9deb18ca
Fix embarrassing typo in the doc chat template! (#26596)
8b03615b
Fix encoder->decoder typo bug in convert_t5x_checkpoint_to_pytorch.py…
ca7912d1
skip flaky hub tests (#26594)
c037b2e3
Update mistral.md to update 404 link (#26590)
f9ab07f9
[Wav2Vec2] Fix tokenizer set lang (#26349)
2d8ee981
add zh translation for installation (#26084)
43bfd093
[ `NougatProcessor`] Fix the default channel (#26608)
b4e66d7a
[`GPTNeoX`] Faster rotary embedding for GPTNeoX (based on llama chang…
253f9a3f
[Falcon] Set `use_cache=False` before creating `presents` which relie…
2ab76c2c
Fix failing tests on `main` due to torch 2.1 (#26607)
54e17a15
Make `ModelOutput` serializable (#26493)
19f0b7dd
[`core`] fix silent bug `keep_in_fp32` modules (#26589)
e6d250e4
#26566 swin2 sr allow in out channels (#26568)
0a3b9d02
Don't close ClearML task if it was created externally (#26614)
9e78c9ac
Fix `transformers-pytorch-gpu` docker build (#26615)
9d206012
[docs] Update to scripts building index.md (#26546)
18fbeec8
Don't install `pytorch-quantization` in Doc Builder docker file (#26622)
75a33d60
Remove unnecessary `view`s of `position_ids` (#26059)
8878eb1b
Fixed inconsistency in several fast tokenizers (#26561)
af38c837
Update tokenization_code_llama_fast.py (#26576)
65aabafe
Remove unnecessary unsqueeze - squeeze in rotary positional embedding…
64845307
Update chat template docs with more tips on writing a template (#26625)
ea52ed9d
fix RoPE t range issue for fp16 (#26602)
87499420
Fix failing `MusicgenTest .test_pipeline_text_to_audio` (#26586)
e840aa67
remove SharedDDP as it is deprecated (#25702)
27597fea
[`LlamaTokenizerFast`] Adds edge cases for the template processor …
9ad815e4
[docstring] Fix docstring for `AlbertConfig` (#26636)
360ea8fc
docs(zh): review and punctuation & space fix (#26627)
897a826d
[DINOv2] Convert more checkpoints (#26177)
2629c8f3
Fixed malapropism error (#26660)
86a4e5a9
fix links in README.md for the GPT, GPT-2, and Llama2 Models (#26640)
8835bff6
Avoid CI OOM (#26639)
740fc6a1
fix typos in idefics.md (#26648)
c7f01bee
[docstring] Fix docstring CLIP configs (#26677)
3763101f
[docstring] Fix docstring for `CLIPImageProcessor` (#26676)
d2f06dff
[docstring] Fix docstring for DonutImageProcessor (#26641)
3257946f
Fix stale bot (#26692)
87b4ade9
[docstring] Fix docstrings for `CLIP` (#26691)
a5e6df82
Control first downsample stride in ResNet (#26374)
592f2eab
Fix Typo: table in deepspeed.md (#26705)
a9862a0f
[docstring] Fix docstring for `LlamaConfig` (#26685)
e8fdd787
fix a typo in flax T5 attention - attention_mask variable is misnamed…
975003ea
Fix source_prefix default value (#26654)
3eceaa36
[JAX] Replace uses of `jnp.array` in types with `jnp.ndarray`. (#26703)
fc639143
Make Whisper Encoder's sinusoidal PE non-trainable by default (#26032)
1e3c9dda
In assisted decoding, pass model_kwargs to model's forward call (fix …
dcc49d8a
Update docs to explain disabling callbacks using report_to (#26155)
9f406392
`Copied from` for test files (#26713)
5334796d
[Assistant Generation] Improve Encoder Decoder (#26701)
da69de17
[docstring] `SwinModel` docstring fix (#26679)
cc44ca80
fix the model card issue as `use_cuda_amp` is no more available (#26731)
69873d52
Fix stale bot for locked issues (#26711)
6ecb2ab6
Fix checkpoint path in `no_trainer` scripts (#26733)
1d6a8474
Update docker files to use `torch==2.1.0` (#26735)
b219ae6b
Revert #20715 (#26734)
e58cbed5
[docstring] Fix docstring for `LlamaTokenizer` and `LlamaTokenizerFas…
aaccf184
[docstring] Fix docstring for `CodeLlamaTokenizer` (#26709)
797a1bab
add japanese documentation (#26138)
9b7668c0
Translated the accelerate.md file of the documentation to Chinese (#2…
e1cec434
Fix doctest for `Blip2ForConditionalGeneration` (#26737)
3bc65505
Add many missing spaces in adjacent strings (#26751)
40ea9ab2
Warnings controlled by logger level (#26527)
ab0ddc99
Fix `PersimmonIntegrationTest` OOM (#26750)
72256bc7
Fix `MistralIntegrationTest` OOM (#26754)
db5e0c32
Fix backward compatibility of Conversation (#26741)
57632bf9
[docs] LLM prompting guide (#26274)
0ebee8b9
[docstring] Fix `UniSpeech`, `UniSpeechSat`, `Wav2Vec2ForCTC` (#26664)
eb734e51
[docstring] Update `GPT2` and `Whisper` (#26642)
b4199c2d
[docstring] Fix docstring for 'BertGenerationConfig' (#26661)
33df09e7
Fix `PerceiverModelIntegrationTest::test_inference_masked_lm` (#26760)
a243cdca
chore: fix typos (#26756)
883ed4b3
Skip `TrainerIntegrationFSDP::test_basic_run_with_cpu_offload` if `to…
3e93dd29
🌐 [i18n-KO] Translated `big_models.md` to Korean (#26245)
7790943c
Update expect outputs of `IdeficsProcessorTest.test_tokenizer_padding…
21da3b24
[docstring] Fix docstring for `RwkvConfig` (#26782)
d085662c
Fix num. of minimal calls to the Hub with peft for pipeline (#26385)
288bf5c1
[docstring] fix docstring `DPRConfig` (#26674)
5bfda28d
[`core`] Fix fa-2 import (#26785)
6df9179c
Disable default system prompt for LLaMA (#26765)
c9785d95
Fix Falcon generation test (#26770)
bdb391e9
Add OWLv2, bis (#26668)
762af3e3
Fixed KeyError for Mistral (#26682)
8e05ad32
[`Flava`] Fix flava doc (#26789)
7cc6f822
Add CLIP resources (#26534)
d6e5b02e
translation brazilian portuguese (#26769)
21dc5859
Fixed typos (#26810)
0dd58d96
[docstring] Fix docstring for `CanineConfig` (#26771)
0e52af4d
Add Japanese translation (#26799)
69a26c7e
[docstring] Fix docstring for `CodeLlamaTokenizerFast` (#26666)
5c081e29
Image-to-Image Task Guide (#26595)
5d997f22
Make fsdp ram efficient loading optional (#26631)
a5f5568d
fix resume_from_checkpoint bug (#26739)
b91cff5a
[OWL-ViT, OWLv2] Add resources (#26822)
570b3f9c
Add LLM doc (#26058)
805d5d21
Llama tokenizer: remove space in template comment (#26788)
3ef71345
Better way to run AMD CI with different flavors (#26634)
12cc1233
[docstring] Fix bert generation tokenizer (#26820)
5c6b83cb
Conversation pipeline fixes (#26795)
14b04b4b
🚨🚨🚨 [`Quantization`] Store the original dtype in the config as a priv…
fd6a0ade
Fix Mistral OOM again (#26847)
b8f1cde9
Chore: Typo fixed in multiple files of docs/source/en/model_doc (#26833)
b3961f72
fix: when window_size is passes as array (#26800)
85e9d644
Update logits_process.py docstrings to clarify penalty and reward cas…
0b8604d0
🚨🚨 Generate: change order of ops in beam sample to avoid nans (#26843)
4b423e60
[`FA2`] Fix flash attention 2 fine-tuning with Falcon (#26852)
41c42f85
🚨 🚨 Raise error when no speaker embeddings in speecht5._generate_spe…
db611aab
[docstring] Fix docstring for LukeConfig (#26858)
51042ae8
Fixed a typo in mistral.md (#26879)
46092f76
Translating `en/internal` folder docs to Japanese 🇯🇵 (#26747)
b002353d
Fix TensorFlow pakage check (#26842)
ef42cb62
Generate: improve docstrings for custom stopping criteria (#26863)
e893b1ef
Bump urllib3 from 1.26.17 to 1.26.18 in /examples/research_projects/v…
6d644d68
Bump urllib3 from 1.26.17 to 1.26.18 in /examples/research_projects/d…
bece55d8
Knowledge distillation for vision guide (#25619)
280c757f
Fix Seq2seqTrainer decoder attention mask (#26841)
34678db4
[`Tokenizer`] Fix slow and fast serialization (#26570)
ef7e9369
Emergency PR to skip conversational tests to fix CI (#26906)
de55ead1
Add default template warning (#26637)
d933818d
Refactor code part in documentation translated to japanese (#26900)
eec5a3a8
[i18n-ZH] Translated fast_tokenizers.md to Chinese (#26910)
732d2a8a
[`FA-2`] Final fix for FA2 dtype (#26846)
5a73316b
Add fuyu model (#26911)
caa0ff0b
[`FA-2`] Revert suggestion that broke FA2 fine-tuning with quantized …
574a5384
[docstring] Fix docstring for `ChineseCLIP` (#26880)
816c2237
[Docs] Make sure important decode and generate method are nicely disp…
734dd96e
Fix and re-enable ConversationalPipeline tests (#26907)
bdbcd5d4
[docstring] Fix docstrings for `CodeGen` (#26821)
ad08137e
Fix license (#26931)
73dc23f7
Pin Keras for now (#26904)
cbd278f0
[`FA-2` / `Mistral`] Supprot fa-2 + right padding + forward (#26912)
bc4bbd9f
Generate: update basic llm tutorial (#26937)
ae4fb846
Corrected modalities description in README_ru.md (#26913)
08a2edfc
[docstring] Fix docstring for speech-to-text config (#26883)
929134bf
fix set_transform link docs (#26856)
9b197669
Fix Fuyu image scaling bug (#26918)
c030fc89
Update README_hd.md (#26872)
224794b0
Added Telugu [te] translations (#26828)
093848d3
fix logit-to-multi-hot conversion in example (#26936)
f71c9ccf
Limit to inferior fsspec version (#27010)
70032949
python falcon doc-string example typo (#26995)
45425660
skip two tests (#27013)
ef978d0a
Nits in Llama2 docstring (#26996)
d33d3131
Change default `max_shard_size` to smaller value (#26942)
50d0cf4f
Add Seamless M4T model (#25693)
cb45f71c
[`NLLB-MoE`] Fix NLLB MoE 4bit inference (#27012)
244a53e0
[`SeamlessM4T`] fix copies with NLLB MoE int8 (#27018)
f9f27b0f
small typos found (#26988)
c0b5ad94
Merge branch 'main' of github.com:huggingface/transformers into add_udop
8f151eb6
fixups
4bdcc24d
more fixups
08685306
fix the tokenizers
6d98a920
remove un-necessary changes
dbbb0990
nits
536e339b
nits
24bc54a5
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub