Commit Graph

889 Commits (c8f69e04a98093708164e7e607f5ad4e5dfb5fb4)

Author SHA1 Message Date
Ross Wightman c8f69e04a9
Merge pull request #1365 from veritable-tech/fix-resize-pos-embed
2 years ago
Ceshine Lee 0b64117592 Take `no_emb_class` into account when calling `resize_pos_embed`
2 years ago
Jasha10 56c3a84db3
Update type hint for `register_notrace_module`
2 years ago
Ross Wightman 1b278136c3 Change models with mean 0,0,0 std 1,1,1 from int to float for consistency as mentioned in #1355
2 years ago
Ross Wightman 909705e7ff Remove some redundant requires_grad=True from nn.Parameter in third party code
2 years ago
Ross Wightman c5e0d1c700 Add dilation support to convnext, allows output_stride=8 and 16 use. Fix #1341
2 years ago
Ross Wightman dc376e3676 Ensure all model entrypoint fn default to `pretrained=False` (a few didn't)
2 years ago
Ross Wightman 23b102064a Add cs3sedarknet_x weights w/ 82.65 @ 288 top1. Add 2 cs3 edgenet models (w/ 3x3-1x1 block), remove aa from cspnet blocks (not needed)
2 years ago
Ross Wightman 0dbd9352ce Add bulk_runner script and updates to benchmark.py and validate.py for better error handling in bulk runs (used for benchmark and validation result runs). Improved batch size decay stepping on retry...
2 years ago
Ross Wightman 92b91af3bb version 0.6.6
2 years ago
Ross Wightman 05313940e2 Add cs3darknet_x, cs3sedarknet_l, and darknetaa53 weights from TPU sessions. Move SE btwn conv1 & conv2 in DarkBlock. Improve SE/attn handling in Csp/DarkNet. Fix leaky_relu bug on older csp models.
2 years ago
nateraw 51cca82aa1 👽 use hf_hub_download instead of cached_download
2 years ago
Ross Wightman 324a4e58b6 disable nvfuser for jit te/legacy modes (for PT 1.12+)
2 years ago
Ross Wightman 2898cf6e41 version 0.6.5 for pypi release
2 years ago
Ross Wightman a45b4bce9a x and xx small edgenext models do benefit from larger test input size
2 years ago
Ross Wightman a8e34051c1 Unbreak gamma remap impacting beit checkpoint load, version bump to 0.6.4
2 years ago
Ross Wightman 1c5cb819f9 bump version to 0.6.3 before merge
2 years ago
Ross Wightman a1cb25066e Add edgnext_small_rw weights trained with swin like recipe. Better than original 'small' but not the recent 'USI' distilled weights.
2 years ago
Ross Wightman 7c7ecd2492 Add --use-train-size flag to force use of train input_size (over test input size) for validation. Default test-time pooling to use train input size (fixes issues).
2 years ago
Ross Wightman ce65a7b29f Update vit_relpos w/ some additional weights, some cleanup to match recent vit updates, more MLP log coord experiments.
2 years ago
Ross Wightman 58621723bd Add CrossStage3 DarkNet (cs3) weights
2 years ago
Ross Wightman 9be0c84715 Change set -> dict w/ None keys for dataset split synonym search, so always consistent if more than 1 exists. Fix #1224
2 years ago
Ross Wightman db0cee9910 Refactor cspnet configuration using dataclasses, update feature extraction for new cs3 variants.
2 years ago
Ross Wightman eca09b8642 Add MobileVitV2 support. Fix #1332. Move GroupNorm1 to common layers (used in poolformer + mobilevitv2). Keep ol custom ConvNeXt LayerNorm2d impl as LayerNormExp2d for reference.
2 years ago
Ross Wightman 06307b8b41 Remove experimental downsample in block support in ConvNeXt. Experiment further before keeping it in.
2 years ago
Ross Wightman bfc0dccb0e Improve image extension handling, add methods to modify / get defaults. Fix #1335 fix #1274.
2 years ago
Ross Wightman 7d4b3807d5 Support DeiT-3 (Revenge of the ViT) checkpoints. Add non-overlapping (w/ class token) pos-embed support to vit.
3 years ago
Ross Wightman d0c5bd5722 Rename cs2->cs3 for darknets. Fix features_only for cs3 darknets.
3 years ago
Ross Wightman d765305821 Remove first_conv for resnetaa50 def
3 years ago
Ross Wightman dd9b8f57c4 Add feature_info to edgenext for features_only support, hopefully fix some fx / test errors
3 years ago
Ross Wightman 377e9bfa21 Add TPU trained darknet53 weights. Add mising pretrain_cfg for some csp/darknet models.
3 years ago
Ross Wightman c170ba3173 Add weights for resnet10t, resnet14t, and resnetaa50 models. Fix #1314
3 years ago
Ross Wightman 188c194b0f Left some experiment stem code in convnext by mistake
3 years ago
Ross Wightman 70d6d2c484 support test_crop_size in data config resolve
3 years ago
Ross Wightman 6064d16a2d Add initial EdgeNeXt import. Significant cleanup / reorg (like ConvNeXt). Fix #1320
3 years ago
Ross Wightman 7a9c6811c9 Add eps arg to LayerNorm2d, add 'tf' (tensorflow) variant of trunc_normal_ that applies scale/shift after sampling (instead of needing to move a/b)
3 years ago
Ross Wightman 82c311d082 Add more experimental darknet and 'cs2' darknet variants (different cross stage setup, closer to newer YOLO backbones) for train trials.
3 years ago
Ross Wightman a050fde5cd Add resnet10t (basic block) and resnet14t (bottleneck) with 1,1,1,1 repeats
3 years ago
Ross Wightman e6d7df40ec no longer a point using kwargs for pretrain_cfg resolve, just pass explicit arg
3 years ago
Ross Wightman 07d0c4ae96 Improve repr for DropPath module
3 years ago
Ross Wightman e27c16b8a0 Remove unecessary code for synbn guard
3 years ago
Ross Wightman 0da3c9ebbf Remove SiLU layer in default args that breaks import on old old PyTorch
3 years ago
Ross Wightman 7d657d2ef4 Improve resolve_pretrained_cfg behaviour when no cfg exists, warn instead of crash. Improve usability ex #1311
3 years ago
Ross Wightman 879df47c0a Support BatchNormAct2d for sync-bn use. Fix #1254
3 years ago
Ross Wightman 7cedc8d474 Follow up to #1256, fix interpolation warning in auto_autoaugment as well
3 years ago
Jakub Kaczmarzyk db64393c0d
use `Image.Resampling` namespace for PIL mapping (#1256)
3 years ago
Ross Wightman 20a1fa63f8 Make dev version 0.6.2.dev0 for pypi pre
3 years ago
Ross Wightman 347308faad Update README.md, version to 0.6.2
3 years ago
Ross Wightman 4b30bae67b Add updated vit_relpos weights, and impl w/ support for official swin-v2 differences for relpos. Add bias control support for MLP layers
3 years ago
Ross Wightman d4c0588012 Remove persistent buffers from Swin-V2. Change SwinV2Cr cos attn + tau/logit_scale to match official, add ckpt convert, init_value zeros resid LN weight by default
3 years ago