Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Aug 11, 2025

• feat: add webp to img_extensions

  lynnmatrix

  authored
  e326b58

  Copy full SHA for e326b58

Commits on Aug 8, 2025

• Never passed torchcompile-mode arg through to compile call

  rwightman

  committed
  954613a

  Copy full SHA for 954613a

Commits on Aug 4, 2025

• Fix #2562, add siglip2 naflex vit encoder weights

  rwightman

  committed
  055a600

  Copy full SHA for 055a600

Commits on Aug 2, 2025

• Update comments for mnv5 pretrained configs

  rwightman

  committed
  b3869b6

  Copy full SHA for b3869b6

• Add support for Gemma 3n MobileNetV5 encoder weight loading

  rwightman

  committed
  6c5e3ce

  Copy full SHA for 6c5e3ce

Commits on Aug 1, 2025

• Dev version 1.0.20.dev0

  rwightman

  committed
  1e1c637

  Copy full SHA for 1e1c637

• Add mobileclip2 encoder weights

  rwightman

  committed
  2ddbe5e

  Copy full SHA for 2ddbe5e

Commits on Jul 26, 2025

• Pass init_values at hieradet_sam2

  

  hassonofer

  authored and

  rwightman

  committed
  435ab7a

  Copy full SHA for 435ab7a

Commits on Jul 24, 2025

• Train script issues. Fix patch_size extraction in naflex mode + distributed. Fix update time calculations after adding naflex mode. Fix #2556

  rwightman

  committed
  55d4b5d

  Copy full SHA for 55d4b5d

• Fix #2555 for Python 3.9 compat. Release 1.0.19

  rwightman

  committed
  d08d5a0

  Copy full SHA for d08d5a0

Commits on Jul 23, 2025

• Fix newer type annotation in layers/typing

  rwightman

  committed
  0e79579

  Copy full SHA for 0e79579

• Release 1.0.18

  rwightman

  committed
  e6ab6bc

  Copy full SHA for e6ab6bc

• Support set_input_size() in EVA models (#2554)

  Show description for 280a90c
  rwightman

  authored
  280a90c

  Copy full SHA for 280a90c

Commits on Jul 21, 2025

• Update README.md

  rwightman

  authored
  19f2bfb

  Copy full SHA for 19f2bfb

Commits on Jul 19, 2025

• Add 8 new Meta Perception Encoder (PE) weight/variants via EVA. Test NaFlexVit support. Fix #2550

  rwightman

  committed
  4a67e13

  Copy full SHA for 4a67e13

• Fix attn pool specific num_heads / mlp_ratio being passed for PE models in NaFlexVit

  rwightman

  committed
  c603c31

  Copy full SHA for c603c31

• Fix PE ViT norm layers in NaFlexVit. Still need a PE Giant fix...

  rwightman

  committed
  6fb3536

  Copy full SHA for 6fb3536

• Fix model patch sizes != 16 in validate.py for naflex_loader use

  rwightman

  committed
  68790f9

  Copy full SHA for 68790f9

• Cleanup eva kwarg handling for naflexvit adaptation, remove debug print

  rwightman

  committed
  6b0842e

  Copy full SHA for 6b0842e

• Shorten comment lines for apply_keep_indices

  rwightman

  committed
  9625dc2

  Copy full SHA for 9625dc2

• Add ROPE support to NaFlexVit (axial and mixed), and support loading of most (all?) EVA based vit models & weights. Fix PatchDropout use w/ ROPE and NaFlexVit. Fix #2549

  rwightman

  committed
  6063f82

  Copy full SHA for 6063f82

Commits on Jul 10, 2025

• Back to dev version 1.0.18.dev0

  rwightman

  committed
  b2034bb

  Copy full SHA for b2034bb

Commits on Jul 9, 2025

• Fix 3 typos in README.md

  

  robin-ede

  authored and

  rwightman

  committed
  59fdbaa

  Copy full SHA for 59fdbaa

• Remove debug print from mnv5 enc module

  rwightman

  committed
  7452176

  Copy full SHA for 7452176

• Fix scripting?

  rwightman

  committed
  2171831

  Copy full SHA for 2171831

• Fix H, W ordering for xy indexing in ROPE, impacts models w/ xy indexing and non-square images

  rwightman

  committed
  33791c8

  Copy full SHA for 33791c8

Commits on Jul 7, 2025

• Update README, bump version to 1.0.17 in prep for release

  rwightman

  committed
  a7c5368

  Copy full SHA for a7c5368

• Fixup casting issues for weights/bias in fp32 norm layers

  rwightman

  committed
  37d9440

  Copy full SHA for 37d9440

• fix: mnv5 with conv_stem bias and GELU approx tanh

  

  RyanMullins

  authored and

  rwightman

  committed
  ea4f940

  Copy full SHA for ea4f940

Commits on Jul 6, 2025

• Add flag to enable float32 computation for normalization (norm + affine) (#2536)

  Show description for 446e8a8
  rwightman

  authored
  446e8a8

  Copy full SHA for 446e8a8

• Merge pull request #2538 from huggingface/mnv5_bias_str_norm

  Show description for 168137b
  rwightman

  authored
  168137b

  Copy full SHA for 168137b

• Add stem_bias option to MNV5. Resolve the norm layer so can pass string args. Fix default input mean/std for mnv5_300m models

  rwightman

  committed
  e1c158e

  Copy full SHA for e1c158e

Commits on Jul 5, 2025

• Merge pull request #2537 from huggingface/no_opt_layer_decay

  Show description for 72b9752
  rwightman

  authored
  72b9752

  Copy full SHA for 72b9752

• Add a min layer-decay scale clamp, and no optimization threshold to exclude groups from optimization

  rwightman

  committed
  83709ae

  Copy full SHA for 83709ae

Commits on Jul 4, 2025

• Merge pull request #2534 from GuillaumeErhard/inference_mode

  Show description for 6239313
  rwightman

  authored
  6239313

  Copy full SHA for 6239313

Pagination

Previous

Next