Commits

Commit history

1. Commits on Oct 22, 2024

   • Fix Korean doc _toctree.yml (#34293)

     Show description for 93352e8
     regisss

     authored
     93352e8View commit details

     Copy full SHA for 93352e8

     Browse repository at this point

   • [docs] Fix GenerationConfig params (#34299)

     Show description for b644178
     stevhliu

     authored
     b644178View commit details

     Copy full SHA for b644178

     Browse repository at this point

   • T5 compile compatibilty (#34089)

     Show description for 73d65e6
     zucchini-nlp

     authored
     73d65e6View commit details

     Copy full SHA for 73d65e6

     Browse repository at this point

   • VLM: add more modularity (#34175)

     Show description for 5077bc0
     zucchini-nlp

     authored
     5077bc0View commit details

     Copy full SHA for 5077bc0

     Browse repository at this point

   • Attn implementation for composite models (#32238)

     Show description for 21d5025
     

     zucchini-nlp

     and

     asartran

     authored
     21d5025View commit details

     Copy full SHA for 21d5025

     Browse repository at this point
2. Commits on Oct 21, 2024

   • Fix method name which changes in tutorial (#34252)

     Show description for 32590b5
     andimarafioti

     authored
     32590b5View commit details

     Copy full SHA for 32590b5

     Browse repository at this point

   • Add a doc section on writing generation prompts (#34248)

     Show description for f701b98
     Rocketknight1

     authored
     f701b98View commit details

     Copy full SHA for f701b98

     Browse repository at this point

   • Add DetrImageProcessorFast (#34063)

     Show description for a412281
     yonigozlan

     authored
     a412281View commit details

     Copy full SHA for a412281

     Browse repository at this point

   • Change Paligemma import logging to work with modular (#34211)

     Show description for 24bdc94
     yonigozlan

     authored
     24bdc94View commit details

     Copy full SHA for 24bdc94

     Browse repository at this point

   • Generation tests: don't rely on main input name (#34228)

     Show description for ca541bd
     zucchini-nlp

     authored
     ca541bdView commit details

     Copy full SHA for ca541bd

     Browse repository at this point
3. Commits on Oct 18, 2024

   • Only cast logits to float when computing loss (#34147)

     Show description for 816f442
     

     Matthew Hoffman

     authored
     816f442View commit details

     Copy full SHA for 816f442

     Browse repository at this point

   • Fix UDOP dtype issue (#34180)

     Show description for e46e3bc
     Rocketknight1

     authored
     e46e3bcView commit details

     Copy full SHA for e46e3bc

     Browse repository at this point

   • add Glm (#33823)

     Show description for 6604764
     Cyrilvallez

     authored
     6604764View commit details

     Copy full SHA for 6604764

     Browse repository at this point

   • Informative 2 (#34154)

     Show description for e95ea47
     

     LysandreJik

     and

     lewtun

     authored
     e95ea47View commit details

     Copy full SHA for e95ea47

     Browse repository at this point

   • Fix broken test decorator `require_torch_up_to_2_accelerators` (#34201)

     Show description for 0437d6c
     byi8220

     authored
     0437d6cView commit details

     Copy full SHA for 0437d6c

     Browse repository at this point

   • BLIP: fix input expansion logic (#34225)

     Show description for 5a5b590
     zucchini-nlp

     authored
     5a5b590View commit details

     Copy full SHA for 5a5b590

     Browse repository at this point
4. Commits on Oct 17, 2024

   • Fix-red-ci (#34230)

     Show description for b54109c
     ArthurZucker

     authored
     b54109cView commit details

     Copy full SHA for b54109c

     Browse repository at this point

   • Enable users to use their own loss functions + deal with prefetching for grad accum (#34198)

     Show description for 6ba31a8
     

     muellerzr

     and

     danielhanchen

     authored
     6ba31a8View commit details

     Copy full SHA for 6ba31a8

     Browse repository at this point

   • Support Llama 3.2 conversion (text models) (#33778)

     Show description for 7a06d07
     

     pcuenca

     and

     osanseviero

     authored
     7a06d07View commit details

     Copy full SHA for 7a06d07

     Browse repository at this point

   • Fix Gradient Accumulation issue (#34191)

     Show description for c1c7e89
     

     ArthurZucker

     and

     kashif

     authored
     c1c7e89View commit details

     Copy full SHA for c1c7e89

     Browse repository at this point

   • Generate: visit non-llm `prepare_inputs_for_generation` (#34199)

     Show description for f51ac9e
     

     gante

     and

     ArthurZucker

     authored
     f51ac9eView commit details

     Copy full SHA for f51ac9e

     Browse repository at this point

   • Fix bus error when using GPT2 on M1 macs (#34031)

     Show description for 1d2c29f
     chanind

     authored
     1d2c29fView commit details

     Copy full SHA for 1d2c29f

     Browse repository at this point

   • Llama3 and Llama2 are ExecuTorch compatible (#34101)

     Show description for 9470c00
     

     guangy10

     and

     Guang Yang

     authored
     9470c00View commit details

     Copy full SHA for 9470c00

     Browse repository at this point

   • removes decord (#33987)

     Show description for 7f50885
     

     vrnvu

     and

     ad

     authored
     7f50885View commit details

     Copy full SHA for 7f50885

     Browse repository at this point

   • Fix for tokenizer.apply_chat_template with continue_final_message=True (#34214)

     Show description for f2846ad
     

     schoennenbeck

     and

     Rocketknight1

     authored
     f2846adView commit details

     Copy full SHA for f2846ad

     Browse repository at this point

   • fix(Wav2Vec2ForCTC): torch export (#34023)

     Show description for b57c7bc
     chrsmcgrr

     authored
     b57c7bcView commit details

     Copy full SHA for b57c7bc

     Browse repository at this point

   • Ping team members for new failed tests in daily CI (#34171)

     Show description for fce1fcf
     ydshieh

     authored
     fce1fcfView commit details

     Copy full SHA for fce1fcf

     Browse repository at this point

   • Fix warning message for fp32_cpu_offloading in bitsandbytes configs (#34079)

     Show description for aa3e35a
     amosyou

     authored
     aa3e35aView commit details

     Copy full SHA for aa3e35a

     Browse repository at this point

   • Update `trainer._get_eval_sampler()` to support `group_by_length` arg (#33514)

     Show description for 6d2b203
     larin92

     authored
     6d2b203View commit details

     Copy full SHA for 6d2b203

     Browse repository at this point
5. Commits on Oct 16, 2024

   • Revert "Fix FSDP resume Initialization issue" (#34193)

     Show description for 3f06f95
     SunMarc

     authored
     3f06f95View commit details

     Copy full SHA for 3f06f95

     Browse repository at this point

   • Avoid using torch's Tensor or PIL's Image in chat template utils if not available (#34165)

     Show description for 3a10c61
     

     RezaRahemtola

     and

     Rocketknight1

     authored
     3a10c61View commit details

     Copy full SHA for 3a10c61

     Browse repository at this point

   • Fix wrong name for llava onevision and qwen2_vl in tokenization auto (#34177)

     Show description for bd5dc10
     yonigozlan

     authored
     bd5dc10View commit details

     Copy full SHA for bd5dc10

     Browse repository at this point

   • Revert `accelerate` error caused by `46d09af` (#34197)

     Show description for cc7d8b8
     steveepreston

     authored
     cc7d8b8View commit details

     Copy full SHA for cc7d8b8

     Browse repository at this point

   • [fix] fix token healing tests and usage errors (#33931)

     Show description for 98bad9c
     alpertunga-bile

     authored
     98bad9cView commit details

     Copy full SHA for 98bad9c

     Browse repository at this point

   • Moshi integration (#33624)

     Show description for 9ba021e
     

     3 people

     authored
     9ba021eView commit details

     Copy full SHA for 9ba021e

     Browse repository at this point