Commits

Breadcrumbs

History for

onmain

User selector

All users

All time

Commit History

Commits on Jun 26, 2025

• [rfc][compile] compile method for DiffusionPipeline (#11705)

  Show description for d93381c
  

  3 peopleauthored
  d93381c

  Copy full SHA for d93381c

Commits on Jun 19, 2025

• make group offloading work with disk/nvme transfers (#11682)

  Show description for 85a916b
  sayakpaul

  authored
  85a916b

  Copy full SHA for 85a916b

Commits on Jun 11, 2025

• [tests] model-level `device_map` clarifications (#11681)

  Show description for 9154566
  sayakpaul

  authored
  9154566

  Copy full SHA for 9154566

Commits on May 13, 2025

• fix: remove `torch_dtype="auto"` option from docstrings (#11513)

  Show description for f8d4a1e
  

  johannaSommer

  and

  DN6

  authored
  f8d4a1e

  Copy full SHA for f8d4a1e

Commits on Apr 8, 2025

• [feat] implement `record_stream` when using CUDA streams during group offloading (#11081)

  Show description for 4b27c4a
  

  sayakpaul

  and

  a-r-r-o-w

  authored
  4b27c4a

  Copy full SHA for 4b27c4a

Commits on Apr 2, 2025

• Revert `save_model` in ModelMixin save_pretrained and use safe_serialization=False in test (#11196)

  hlky

  authored
  da857be

  Copy full SHA for da857be

• allow models to run with a user-provided dtype map instead of a single dtype (#10301)

  Show description for d8c617c
  

  hlky

  and

  sayakpaul

  authored
  d8c617c

  Copy full SHA for d8c617c

Commits on Mar 21, 2025

• Don't override `torch_dtype` and don't use when `quantization_config` is set (#11039)

  Show description for a7d53a5
  

  hlky

  and

  sayakpaul

  authored
  a7d53a5

  Copy full SHA for a7d53a5

• [core] FasterCache (#10163)

  Show description for 844221a
  

  a-r-r-o-w

  and

  github-actions[bot]

  authored
  844221a

  Copy full SHA for 844221a

Commits on Mar 20, 2025

• Provide option to reduce CPU RAM usage in Group Offload (#11106)

  Show description for 2c1ed50
  DN6

  authored
  2c1ed50

  Copy full SHA for 2c1ed50

Commits on Mar 2, 2025

• [Alibaba Wan Team] continue on #10921 Wan2.1 (#10922)

  Show description for 2d8a41c
  

  4 peopleauthored
  2d8a41c

  Copy full SHA for 2d8a41c

Commits on Feb 24, 2025

• Fix `torch_dtype` in Kolors text encoder with `transformers` v4.49 (#10816)

  Show description for 6f74ef5
  hlky

  authored
  6f74ef5

  Copy full SHA for 6f74ef5

Commits on Feb 20, 2025

• Remove print statements (#10836)

  Show description for f8b54cf
  a-r-r-o-w

  authored
  f8b54cf

  Copy full SHA for f8b54cf

Commits on Feb 19, 2025

• [FEAT] Model loading refactor (#10604)

  Show description for f5929e0
  

  4 peopleauthored
  f5929e0

  Copy full SHA for f5929e0

Commits on Feb 14, 2025

• Module Group Offloading (#10503)

  Show description for 9a147b8
  

  a-r-r-o-w

  and

  stevhliu

  authored
  9a147b8

  Copy full SHA for 9a147b8

Commits on Feb 13, 2025

• make tensors contiguous before passing to safetensors (#10761)

  Show description for 97abdd2
  faaany

  authored
  97abdd2

  Copy full SHA for 97abdd2

Commits on Feb 10, 2025

• Add `Self` type hint to `ModelMixin`'s `from_pretrained` (#10742)

  hlky

  authored
  7fb481f

  Copy full SHA for 7fb481f

Commits on Jan 28, 2025

• Refactor gradient checkpointing (#10611)

  Show description for c4d4ac2
  a-r-r-o-w

  authored
  c4d4ac2

  Copy full SHA for c4d4ac2

Commits on Jan 22, 2025

• [core] Layerwise Upcasting (#10347)

  Show description for beacaa5
  

  3 peopleauthored
  beacaa5

  Copy full SHA for beacaa5

Commits on Jan 21, 2025

• [chore] change licensing to 2025 from 2024. (#10615)

  Show description for 4ace7d0
  sayakpaul

  authored
  4ace7d0

  Copy full SHA for 4ace7d0

Commits on Jan 16, 2025

• implementing flux on TPUs with ptxla (#10515)

  Show description for aeac0a0
  

  4 peopleauthored
  aeac0a0

  Copy full SHA for aeac0a0

• Move buffers to device (#10523)

  Show description for 0b065c0
  hlky

  authored
  0b065c0

  Copy full SHA for 0b065c0

Commits on Jan 14, 2025

• [FEAT] DDUF format (#10037)

  Show description for fbff43a
  

  5 peopleauthored
  fbff43a

  Copy full SHA for fbff43a

Commits on Jan 10, 2025

• Add a `disable_mmap` option to the `from_single_file` loader to improve load performance on network mounts (#10305)

  Show description for 52c05bd
  

  8 peopleauthored
  52c05bd

  Copy full SHA for 52c05bd

Commits on Jan 8, 2025

• Fix compatibility with pipeline when loading model with device_map on single gpu (#10390)

  Show description for e2deb82
  

  SunMarc

  and

  sayakpaul

  authored
  e2deb82

  Copy full SHA for e2deb82

• Add `_no_split_modules` to some models (#10308)

  Show description for 71ad16b
  

  a-r-r-o-w

  and

  DN6

  authored
  71ad16b

  Copy full SHA for 71ad16b

Commits on Dec 25, 2024

• Fix TorchAO related bugs; revert device_map changes (#10371)

  Show description for cd991d1
  

  a-r-r-o-w

  and

  sayakpaul

  authored
  cd991d1

  Copy full SHA for cd991d1

Commits on Dec 23, 2024

• update `get_parameter_dtype` (#10342)

  Show description for 7c2f0af
  yiyixuxu

  authored
  7c2f0af

  Copy full SHA for 7c2f0af

Commits on Dec 20, 2024

• Add support for sharded models when TorchAO quantization is enabled (#10256)

  Show description for 41ba8c0
  a-r-r-o-w

  authored
  41ba8c0

  Copy full SHA for 41ba8c0

Commits on Dec 17, 2024

• [Single File] Add GGUF support (#9964)

  Show description for e24941b
  

  3 peopleauthored
  e24941b

  Copy full SHA for e24941b

Commits on Dec 16, 2024

• [core] TorchAO Quantizer (#10009)

  Show description for 9f00c61
  

  3 peopleauthored
  9f00c61

  Copy full SHA for 9f00c61

Commits on Dec 6, 2024

• Update ptxla training (#9864)

  Show description for 3cb7b86
  

  6 peopleauthored
  3cb7b86

  Copy full SHA for 3cb7b86

Commits on Dec 5, 2024

• Use torch.device instead of current device index for BnB quantizer (#10069)

  Show description for 98d0cd5
  

  a-r-r-o-w

  and

  sayakpaul

  authored
  98d0cd5

  Copy full SHA for 98d0cd5

Commits on Dec 3, 2024

• Let server decide default repo visibility (#10047)

  Wauplin

  authored
  0763a7e

  Copy full SHA for 0763a7e

Commits on Oct 21, 2024

• [Quantization] Add quantization support for `bitsandbytes` (#9213)

  Show description for b821f00
  

  4 peopleauthored
  b821f00

  Copy full SHA for b821f00