Commits

Breadcrumbs

History for

on01cb46a

User selector

All users

Datepicker

All time

Commit History

Commits on Aug 7, 2021

• Add gc_efficientnetv2_rw_t weights (global context instead of SE attn). Add TF XL weights even though the fine-tuned ones don't validate that well. Change default arg for GlobalContext to use scal …

  Show description for 01cb46a
  rwightman

  committed
  01cb46a

  Copy full SHA for 01cb46a

Commits on Jul 22, 2021

• Add EfficientNetV2 XL model defs

  rwightman

  committed
  d3f7440

  Copy full SHA for d3f7440

Commits on Jul 11, 2021

• Fix gc effv2 model cfg name

  rwightman

  committed
  f8039c7

  Copy full SHA for f8039c7

Commits on Jul 9, 2021

• Add efficientnetv2_rw_t defs w/ weights, and gc variant, as well as gcresnet26ts for experiments. Version 0.4.13

  rwightman

  committed
  392368e

  Copy full SHA for 392368e

Commits on May 31, 2021

• Add non-local and BAT attention. Merge attn and self-attn factories into one. Add attention references to README. Add mlp 'mode' to ECA.

  rwightman

  committed
  307a935

  Copy full SHA for 307a935

Commits on May 30, 2021

• Able to use other attn layer in EfficientNet now. Create test ECA + GC B0 configs. Make ECA more configurable.

  rwightman

  committed
  8bf63b6

  Copy full SHA for 8bf63b6

Commits on May 22, 2021

• Add efficientnetv2_rw_m weights trained in PyTorch. 84.8 top-1 @ 416 test. 53M params.

  rwightman

  committed
  23c18a3

  Copy full SHA for 23c18a3

• Prep for effcientnetv2_rw_m model weights that started training before official release..

  rwightman

  committed
  c2ba229

  Copy full SHA for c2ba229

Commits on May 15, 2021

• Change 21k model naming from _21k to _in21k for consistency with existing 21k models.

  rwightman

  committed
  7077f16

  Copy full SHA for 7077f16

Commits on May 14, 2021

• Update README, tweak fine-tune effv2 model names.

  rwightman

  committed
  328249f

  Copy full SHA for 328249f

• EfficientNetV2 official impl w/ weights ported from TF. Cleanup/refactor of related EfficientNet classes and models.

  rwightman

  committed
  c4f482a

  Copy full SHA for c4f482a

Commits on Apr 19, 2021

• Add updated EfficientNet-V2S weights, 83.8 @ 384x384 test. Add PyTorch trained EfficientNet-B4 weights, 83.4 @ 384x384 test. Tweak non TF EfficientNet B1-B4 train/test res scaling.

  rwightman

  committed
  c1cf971

  Copy full SHA for c1cf971

Commits on Apr 17, 2021

• Test input size for efficientnet_v2s was wrong in last results run

  rwightman

  committed
  e8a64fb

  Copy full SHA for e8a64fb

Commits on Apr 12, 2021

• Add prelim efficientnet_v2s weights from 224x224 train, eval 83.3 @ 288. Add eca_nfnet_l1 weights, train at 256, eval 84 @ 320.

  rwightman

  committed
  21812d3

  Copy full SHA for 21812d3

Commits on Apr 3, 2021

• Update some comments re preliminary EfficientNet-V2 assumptions

  rwightman

  committed
  fb896c0

  Copy full SHA for fb896c0

Commits on Apr 2, 2021

• EfficientNet-V2S preliminary model def (for experimentation)

  rwightman

  committed
  de9dff9

  Copy full SHA for de9dff9

Commits on Mar 17, 2021

• Merge pull request #502 from khatami-mehrdad/efficientnet_dg

  Show description for 5e2e4e7
  rwightman

  authored
  5e2e4e7

  Copy full SHA for 5e2e4e7

• adding efficientnet_el, efficientnet_es_prunedefficientnet_es_pruned and efficientnet_el_pruned pretrained models to timm/models

  

  Mehrdad

  committed
  97f49d1

  Copy full SHA for 97f49d1

• Support for huggingface hub via create_model and default_cfgs.

  Show description for d584e7f
  rwightman

  committed
  d584e7f

  Copy full SHA for d584e7f

Commits on Dec 3, 2020

• Bug in last mod for features_only default_cfg

  rwightman

  committed
  cd72e66

  Copy full SHA for cd72e66

• Add default_cfg back to models wrapped in feature extraction module as per discussion in #294.

  rwightman

  committed
  867a0e5

  Copy full SHA for 867a0e5

Commits on Sep 24, 2020

• Add EfficientNet-EdgeTPU-M (efficientnet_em) model trained natively in PyTorch. More sotabench fiddling.

  rwightman

  committed
  9c40653

  Copy full SHA for 9c40653

Commits on Aug 11, 2020

• Add new EfficientNet-B3 and RegNetY-3.2GF weights, both just over 82 top-1

  rwightman

  committed
  512b2dd

  Copy full SHA for 512b2dd

Commits on Aug 4, 2020

• More uniform treatment of classifiers across all models, reduce code duplication.

  rwightman

  committed
  b1f1a54

  Copy full SHA for b1f1a54

Commits on Jul 29, 2020

• Add EfficientNet-Lite0 weights trained with this code by @hal-314, 75.484 top-1

  rwightman

  committed
  ec4976f

  Copy full SHA for ec4976f

Commits on Jul 27, 2020

• Fix some attributions, add copyrights to some file docstrings

  rwightman

  committed
  6c17d57

  Copy full SHA for 6c17d57

Commits on Jul 25, 2020

• Cleanup FeatureInfo getters, add TF models sourced Xception41/65/71 weights

  rwightman

  committed
  08016e8

  Copy full SHA for 08016e8

Commits on Jul 21, 2020

• Cleanup EfficientNet/MobileNetV3 feature extraction a bit, only two tap locations now, small mobilenetv3 models work

  rwightman

  committed
  c146b54

  Copy full SHA for c146b54

Commits on Jul 19, 2020

• Move FeatureHooks into features.py, switch EfficientNet, MobileNetV3 to use build model helper

  rwightman

  committed
  6eec3fb

  Copy full SHA for 6eec3fb

Commits on Jul 18, 2020

• Lots of changes to model creation helpers, close to finalizing feature extraction / interfaces

  rwightman

  committed
  3b9004b

  Copy full SHA for 3b9004b

Commits on Jun 30, 2020

• Working on feature extraction, interfaces refined, a number of models working, some in progress.

  rwightman

  committed
  d23a269

  Copy full SHA for d23a269

Commits on Jun 3, 2020

• Add missing feature_info() on MobileNetV3, make hook feature output order/type consistent with bottleneck (list, decreasing fmap size)

  rwightman

  committed
  7be2995

  Copy full SHA for 7be2995

Commits on Jun 2, 2020

• Monster commit, activation refactor, VoVNet, norm_act improvements, more

  Show description for eb76536
  rwightman

  committed
  eb76536

  Copy full SHA for eb76536

Commits on May 6, 2020

• Replace all None by nn.Identity() in all models reset_classifier when False-values num_classes is given.

  Show description for a7ebe09
  Animatory

  committed
  a7ebe09

  Copy full SHA for a7ebe09

Commits on May 3, 2020

• added efficientnet pruned weights

  

  AFLALO, Jonathan Isaac

  committed
  9c15d57

  Copy full SHA for 9c15d57

Pagination

Previous

Next