Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Aug 20, 2024

• Improve config handling and add a zoo (huggingface#3029)

  Show description for 1a6af0b
  muellerzr

  authored
  1a6af0b

  Copy full SHA for 1a6af0b

• Add end_training/destroy_pg to everything and unpin numpy (huggingface#3030)

  Show description for 52fae09
  muellerzr

  authored
  52fae09

  Copy full SHA for 52fae09

Commits on Aug 19, 2024

• Fix torch version check (huggingface#3024)

  Show description for 7ffe766
  muellerzr

  authored
  7ffe766

  Copy full SHA for 7ffe766

• Set correct NPU backend and distributed_type when using transfer_to_npu (huggingface#3021)

  Show description for 5536a3a
  

  ArthurinRUC

  and

  yangyuanhang7

  authored
  5536a3a

  Copy full SHA for 5536a3a

• Tweak defaults for quantized-typed FP8 TE weights (huggingface#3018)

  Show description for 7ec8eab
  muellerzr

  authored
  7ec8eab

  Copy full SHA for 7ec8eab

Commits on Aug 15, 2024

• destroy process group in end_training (huggingface#3012)

  Show description for 589fddd
  SunMarc

  authored
  589fddd

  Copy full SHA for 589fddd

• Wrong import check for TE (huggingface#3016)

  muellerzr

  authored
  99c69aa

  Copy full SHA for 99c69aa

• fix default value for rank size in cpu threads_per_process assignment logic (huggingface#3009)

  Show description for 00785cd
  rbrugaro

  authored
  00785cd

  Copy full SHA for 00785cd

Commits on Aug 14, 2024

• Enable FSDP & Deepspeed + FP8 (huggingface#2983)

  Show description for a452327
  muellerzr

  authored
  a452327

  Copy full SHA for a452327

Commits on Aug 13, 2024

• Fix find_tied_params for models with shared layers (huggingface#2986)

  Show description for 851cf34
  

  qubvel

  and

  muellerzr

  authored
  851cf34

  Copy full SHA for 851cf34

Commits on Aug 12, 2024

• update version to 0.34.dev0 (huggingface#3007)

  SunMarc

  authored
  cd5698b

  Copy full SHA for cd5698b

• Add small util to enable FSDP offloading quickly (huggingface#3006)

  Show description for 90d5023
  muellerzr

  authored
  90d5023

  Copy full SHA for 90d5023

• Make env variables optional for FSDP (huggingface#2998)

  Show description for 3bde615
  muellerzr

  authored
  3bde615

  Copy full SHA for 3bde615

Commits on Aug 9, 2024

• Fix deepspeed tests (huggingface#3003)

  Show description for dc3b5ad
  muellerzr

  authored
  dc3b5ad

  Copy full SHA for dc3b5ad

• clear memory after offload (huggingface#2994)

  SunMarc

  authored
  12a5bef

  Copy full SHA for 12a5bef

Commits on Aug 8, 2024

• Support skip_first_batches for XLA (huggingface#2966)

  Show description for 79ca85c
  yitongh

  authored
  79ca85c

  Copy full SHA for 79ca85c

Commits on Aug 7, 2024

• Fix typo on warning str: "meta device device" -> "meta device" (huggingface#2997)

  HeAndres

  authored
  13d93c4

  Copy full SHA for 13d93c4

Commits on Aug 6, 2024

• Explicit check for step when loading the state (huggingface#2992)

  Show description for d982751
  muellerzr

  authored
  d982751

  Copy full SHA for d982751

• Fix gated test (huggingface#2993)

  Show description for 95edc68
  muellerzr

  authored
  95edc68

  Copy full SHA for 95edc68

Commits on Aug 1, 2024

• Fix bug of clip_grad_norm_ for xla fsdp (huggingface#2941)

  Show description for 288accc
  hanwen-sun

  authored
  288accc

  Copy full SHA for 288accc

• remove .md to allow proper linking (huggingface#2977)

  nbroad1881

  authored
  83b0610

  Copy full SHA for 83b0610

Commits on Jul 31, 2024

• add MLU devices for rng state saving and loading. (huggingface#2940)

  Show description for 386f7d2
  

  huismiling

  and

  muellerzr

  authored
  386f7d2

  Copy full SHA for 386f7d2

• chore: Update runs-on configuration for CI workflows (huggingface#2981)

  Show description for 308a8e9
  XciD

  authored
  308a8e9

  Copy full SHA for 308a8e9

• Enable Unwrapping for Model State Dicts (FSDP) (huggingface#2959)

  Show description for f35cbd1
  alex-jw-brooks

  authored
  f35cbd1

  Copy full SHA for f35cbd1

• Fix torchvision to be compatible with torch version in CI (huggingface#2982)

  Show description for a14260c
  SunMarc

  authored
  a14260c

  Copy full SHA for a14260c

Commits on Jul 29, 2024

• Require safetensors>=0.4.3 (huggingface#2957)

  byi8220

  authored
  32f368e

  Copy full SHA for 32f368e

Commits on Jul 22, 2024

• feat(ci): add pip caching in CI (huggingface#2952)

  SauravMaheshkar

  authored
  415eddf

  Copy full SHA for 415eddf

• Properly handle Params4bit in set_module_tensor_to_device (huggingface#2934)

  Show description for 2308576
  matthewdouglas

  authored
  2308576

  Copy full SHA for 2308576

Commits on Jul 20, 2024

• Add torch.float8_e4m3fn format dtype_byte_size (huggingface#2945)

  Show description for a5a3e57
  SunMarc

  authored
  a5a3e57

  Copy full SHA for a5a3e57

Commits on Jul 18, 2024

• delete CCL env var setting (huggingface#2927)

  Show description for 0af1d8b
  Liangliang-Ma

  authored
  0af1d8b

  Copy full SHA for 0af1d8b

Commits on Jul 16, 2024

• Improve test reliability for Accelerator.free_memory() (huggingface#2935)

  matthewdouglas

  authored
  d16d737

  Copy full SHA for d16d737

• Consider pynvml available when installed through the nvidia-ml-py distribution (huggingface#2936)

  matthewdouglas

  authored
  7a5c231

  Copy full SHA for 7a5c231

Commits on Jul 15, 2024

• Fix import test (huggingface#2931)

  Show description for 4f02bb7
  muellerzr

  authored
  4f02bb7

  Copy full SHA for 4f02bb7

• Hotfix PyTorch Version Installation in CI Workflow for Minimum Version Matrix (huggingface#2889)

  Show description for 709fd1e
  yhna940

  authored
  709fd1e

  Copy full SHA for 709fd1e

• Correct loading of models with shared tensors when using accelerator.load_state() (huggingface#2875)

  Show description for f4f1260
  jkuntzer

  authored
  f4f1260

  Copy full SHA for f4f1260

Pagination

Previous

Next