Migrations
/
pytorch-image-models
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Commit Graph

  • 26c400bf1e
         Merge pull request #1498 from nateraw/docs-deploy-from-main Ross Wightman 2022-10-16 20:40:00 -0700
  • cf549c28b1 🎨 deploy docs from main not master #1498 nateraw 2022-10-16 15:51:29 -0400
  • 3e57377f7a
         Update README.md Ross Wightman 2022-10-15 09:33:34 -0700
  • 6635bc3f7d
         Merge pull request #1479 from rwightman/script_cleanup Ross Wightman 2022-10-15 09:29:39 -0700
  • 0e6023f032
         Merge pull request #1381 from ChristophReich1996/master Ross Wightman 2022-10-14 18:34:33 -0700
  • b4ea69c9ce
         Merge pull request #1414 from dedeswim/bits_and_tpu Ross Wightman 2022-10-14 16:31:01 -0700
  • 66f4af7090 Merge remote-tracking branch 'origin/master' into script_cleanup #1479 script_cleanup Ross Wightman 2022-10-14 15:54:00 -0700
  • eb333d6641 Update validate.py to use updated amp args for impl/dtype Ross Wightman 2022-10-14 15:51:20 -0700
  • d3961536c9 comment some debug logs for WDS dataset Ross Wightman 2022-10-14 15:39:00 -0700
  • e9dccc918c Rename dataset/parsers -> dataset/readers, create_parser to create_reader, etc Ross Wightman 2022-10-14 15:14:38 -0700
  • 8c28363dc9 Version 0.7.dev0 for master Ross Wightman 2022-10-14 09:38:02 -0700
  • 1ae6ba62cd
         Merge pull request #1494 from nateraw/add-dev-suffix-to-tag Ross Wightman 2022-10-13 14:23:19 -0700
  • 30bafd7347 🔖 add dev suffix to version tag #1494 nateraw 2022-10-13 17:08:33 -0400
  • 94aed25224
         Merge pull request #1493 from nateraw/fix-hf-docs-build Ross Wightman 2022-10-12 10:08:19 -0700
  • edceac4aaf 💚 fix ci build #1493 nateraw 2022-10-12 13:02:33 -0400
  • 2f041b84da
         Merge pull request #1473 from nateraw/hfdocs Ross Wightman 2022-10-12 09:40:56 -0700
  • f67a7ee8bd Set num_workers in Iterable WDS/TFDS datasets early so sample estimate is correct Ross Wightman 2022-10-11 15:11:18 -0700
  • bb49d02aae
         Update README.md Ross Wightman 2022-10-11 13:13:06 -0700
  • b944546200
         Update README.md Ross Wightman 2022-10-10 22:13:05 -0700
  • cea8df3d0c Version 0.6.12 Ross Wightman 2022-10-10 21:49:52 -0700
  • 9914f744dc Add more maxxvit weights includ ConvNeXt conv block based experiments. Ross Wightman 2022-10-10 21:49:18 -0700
  • 285771972e Change --amp flags, no more --apex-amp and --native-amp, add --amp-impl to select apex, and --amp-dtype to allow bfloat16 AMP dtype Ross Wightman 2022-10-07 15:27:25 -0700
  • b1b024dfed Scheduler update, add v2 factory method, support scheduling on updates instead of just epochs. Add LR to summary csv. Add lr_base scaling calculations to train script. Fix #1168 Ross Wightman 2022-10-07 10:43:04 -0700
  • 4f18d6dc5f Fix logs in WDS parser Ross Wightman 2022-10-07 10:06:17 -0700
  • af0c644c5d 👷 remove --not_python_module flag #1473 nateraw 2022-10-06 14:36:39 -0400
  • df29d46872 ♻️ old doc links -> new doc links nateraw 2022-10-06 14:00:50 -0400
  • 8ef9c48cc5 👷 use main branch of doc-builder thats fixed now nateraw 2022-10-06 13:54:57 -0400
  • 9343f6e431 👷 update ci nateraw 2022-10-05 13:58:17 -0400
  • 5f9eef712c 👷 bump ci nateraw 2022-10-05 13:05:17 -0400
  • 09a77700f6 👷 bump back. it was actions outage not me nateraw 2022-10-05 12:50:02 -0400
  • b8e3ffd498 👷 revert ci workflow to main nateraw 2022-10-05 12:40:44 -0400
  • d56f9f7a23 🐛 fix link nateraw 2022-10-05 12:39:34 -0400
  • 6153816250 🚧 test doc-builder branch to fix build here nateraw 2022-10-05 12:27:17 -0400
  • 145c4d8e46
         Merge pull request #1483 from MohamedAliRashad/patch-3 Ross Wightman 2022-10-05 08:59:37 -0700
  • 8fda68aff6
         Fix repo id bug #1483 Mohamed Rashad 2022-10-05 16:26:06 +0200
  • 2f1abc2cf9 👷 add build_documentation workflow nateraw 2022-10-04 16:38:26 -0400
  • 3a429d04ee 🚑 supply --not_python_module for now nateraw 2022-10-03 17:07:24 -0400
  • d4d915caf3 🚧 update path_to_docs in pr doc builder workflow nateraw 2022-10-03 17:03:11 -0400
  • 96109a909b 🚧 update docs path nateraw 2022-10-03 13:35:44 -0400
  • 6c3d02a7e5 🚧 update inputs to build_pr_documentation workflow nateraw 2022-10-03 13:25:21 -0400
  • 25059001a9 🚧 add repo_owner nateraw 2022-10-03 13:12:09 -0400
  • b8c8550841 Data improvements. Improve train support for in_chans != 3. Add wds dataset support from bits_and_tpu branch w/ fixes and tweaks. TFDS tweaks. Ross Wightman 2022-09-29 16:42:58 -0700
  • d4ea5c7d7d
         Merge pull request #1476 from Afafard/class_dict v0.6.11 Ross Wightman 2022-09-27 13:26:34 -0700
  • 7327792f39 update to support pickle based dictionaries #1476 Alex Fafard 2022-09-27 11:13:48 -0400
  • 1199c5a1a4 clip_laion2b models need 1e-5 eps for LayerNorm Ross Wightman 2022-09-25 10:36:54 -0700
  • e4c99d2bd6 📝 add hfdocs documentation nateraw 2022-09-23 17:51:45 -0700
  • 87939e6fab Refactor device handling in scripts, distributed init to be less 'cuda' centric. More device args passed through where needed. Ross Wightman 2022-09-23 16:08:59 -0700
  • c88947ad3d Add initial Hugging Face Datasets parser impl. Ross Wightman 2022-09-23 16:08:19 -0700
  • e858912e0c Add brute-force checkpoint remapping option Ross Wightman 2022-09-23 16:07:03 -0700
  • b293dfa595 Add CL SE module Ross Wightman 2022-09-23 16:06:09 -0700
  • 2a296412be Add Adan optimizer Ross Wightman 2022-09-23 16:05:52 -0700
  • 5dc4343308 version 0.6.11 Ross Wightman 2022-09-23 13:54:56 -0700
  • a383ef99f5 Make huggingface_hub necessary if it's the only source for a pretrained weight Ross Wightman 2022-09-23 13:54:21 -0700
  • d199f6651d
         Merge pull request #1467 from rwightman/clip_laion2b Ross Wightman 2022-09-23 13:18:16 -0700
  • 81f1a874d1 add link to PP-LCNet #1471 user3984 2022-09-21 07:30:02 +0000
  • 33e30f8c8b Remove layer-decay print #1467 clip_laion2b Ross Wightman 2022-09-18 21:33:03 -0700
  • e069249a2d Add hf hub entries for laion2b clip models, add huggingface_hub dependency, update some setup/reqs, torch >= 1.7 Ross Wightman 2022-09-16 21:39:05 -0700
  • 9d65557be3 Fix errant import Ross Wightman 2022-09-15 17:47:23 -0700
  • 9709dbaaa9 Adding support for fine-tune CLIP LAION-2B image tower weights for B/32, L/14, H/14 and g/14. Still WIP Ross Wightman 2022-09-15 17:25:59 -0700
  • a520da9b49 Update tresnet features_info for v2 Ross Wightman 2022-09-13 20:54:54 -0700
  • c8ab747bf4 BEiT-V2 checkpoints didn't remove 'module' from weights, adapt checkpoint filter Ross Wightman 2022-09-13 17:56:49 -0700
  • 73049dc2aa Fix type in dla weight update Ross Wightman 2022-09-13 17:52:45 -0700
  • 3599c7e6a4 version 0.6.10 Ross Wightman 2022-09-13 16:37:02 -0700
  • e11efa872d Update a bunch of weights with external links to timm release assets. Fixes issue with *aliyuncs.com returning forbidden. Did pickle scan / verify and re-hash. Add TresNet-V2-L weights. Ross Wightman 2022-09-13 16:35:26 -0700
  • 744e010a2b
         Merge 4bbd9a0977 into fa8c84eede #1253 Mohamed Rashad 2022-09-08 21:27:36 +0100
  • 64a558b536
         Merge 38594ef7fd into fa8c84eede #804 Ross Wightman 2022-09-07 23:55:10 +0000
  • 38594ef7fd Merge remote-tracking branch 'origin/master' into bits_and_tpu Ross Wightman 2022-09-07 16:54:29 -0700
  • fa8c84eede Update maxvit_tiny_256 weight to better iter, add coatnet / maxvit / maxxvit model defs for future runs Ross Wightman 2022-09-07 12:37:37 -0700
  • de40f66536 Update README.md Ross Wightman 2022-09-07 10:40:58 -0700
  • c1b3cea19d Add maxvit_rmlp_tiny_rw_256 model def and weights w/ 84.2 top-1 @ 256, 84.8 @ 320 Ross Wightman 2022-09-07 10:27:11 -0700
  • da6f8f5a40 Fix beitv2 tests Ross Wightman 2022-09-07 08:09:47 -0700
  • 914544fc81 Add beitv2 224x224 checkpoints from https://github.com/microsoft/unilm/tree/master/beit2 Ross Wightman 2022-09-06 20:25:18 -0700
  • dc90816f26 Add `maxvit_tiny_rw_224` weights 83.5 @ 224 and `maxvit_rmlp_pico_rw_256` relpos weights, 80.5 @ 256, 81.3 @ 320 Ross Wightman 2022-09-06 16:14:41 -0700
  • f489f02ad1 Make gcvit window size ratio based to improve resolution changing support #1449. Change default init to original. Ross Wightman 2022-09-06 15:10:52 -0700
  • 87bfb055c0 Merge remote-tracking branch 'origin/master' into bits_and_tpu Ross Wightman 2022-09-02 11:28:52 -0700
  • c45c6ee8e4 Update README.md Ross Wightman 2022-08-29 15:52:24 -0700
  • 7f1b223c02 Add maxvit_rmlp_nano_rw_256 model def & weights, make window/grid size dynamic wrt img_size by default Ross Wightman 2022-08-29 15:49:32 -0700
  • e6a4361306 pretrained_cfg entry for mvitv2_small_cls Ross Wightman 2022-08-28 15:27:01 -0700
  • f66e5f0e35 Fix class token support in MViT-V2, add small_class variant to ensure it's tested. Fix #1443 Ross Wightman 2022-08-28 15:24:04 -0700
  • b94b7cea65 Missed GCVit in README paper links Ross Wightman 2022-08-28 15:22:28 -0700
  • f1d2160d85 Update a few maxxvit comments, rename PartitionAttention -> PartitionAttenionCl for consistency with other blocks Ross Wightman 2022-08-26 12:53:49 -0700
  • eca6f0a25c Fix syntax error (extra dataclass comma) in maxxvit.py Ross Wightman 2022-08-26 11:29:09 -0700
  • 4f72bae43b
         Merge pull request #1415 from rwightman/more_vit Ross Wightman 2022-08-26 10:00:42 -0700
  • 1dced6066c
         Merge branch 'rwightman:bits_and_tpu' into bits_and_tpu #1414 Edoardo Debenedetti 2022-08-26 15:47:09 +0200
  • 151fc8a14f add next-vit #1439 XiaXin-Aloys 2022-08-26 19:35:21 +0800
  • ff6a919cf5 Add --fast-norm arg to benchmark.py, train.py, validate.py #1415 more_vit Ross Wightman 2022-08-25 17:00:54 -0700
  • 769ab4b98a Clean up no_grad for trunc normal weight inits Ross Wightman 2022-08-25 16:29:52 -0700
  • 48e1df8b37 Add norm/norm_act header comments Ross Wightman 2022-08-25 16:29:34 -0700
  • 99ee61e245 Add T/G legend to README.md maxvit list Ross Wightman 2022-08-25 15:58:57 -0700
  • a54008bd97 Update README.md for merge Ross Wightman 2022-08-25 15:56:56 -0700
  • 7c2660576d Tweak init for convnext block using maxxvit/coatnext. Ross Wightman 2022-08-25 15:30:59 -0700
  • 1d8d6f6072 Fix two default args in DenseNet blocks... fix #1427 Ross Wightman 2022-08-25 15:00:35 -0700
  • 527f9a4cb2 Updated to correct maxvit_nano weights... Ross Wightman 2022-08-24 12:42:11 -0700
  • 2a5b5b2a7b
         Update feature_request.md Ross Wightman 2022-08-24 12:24:32 -0700
  • e018253acc
         Update config.yml Ross Wightman 2022-08-24 12:21:03 -0700
  • 995e2691d6
         Update config.yml Ross Wightman 2022-08-24 12:20:26 -0700
  • f07dfe010a Merge remote-tracking branch 'origin/more_vit' into bits_and_tpu Ross Wightman 2022-08-24 12:11:34 -0700
  • b2e8426fca Make k=stride=2 ('avg2') pooling default for coatnet/maxvit. Add weight links. Rename 'combined' partition to 'parallel'. v0.1-weights-maxx Ross Wightman 2022-08-24 11:01:20 -0700
  • 837c68263b For ConvNeXt, use timm internal LayerNorm for fast_norm in non conv_mlp mode Ross Wightman 2022-08-23 15:17:12 -0700
  • cac0a4570a More test fixes, pool size for 256x256 maxvit models Ross Wightman 2022-08-23 13:38:26 -0700