1417 Commits

Author SHA1 Message Date
Ross Wightman
5c5ae8d401 Fix #2132, remove use of _C.set_grad_enable. Line endings were messed up too 2024-04-09 09:00:23 -07:00
fzyzcjy
b44e4e45a2 more 2024-04-02 10:25:30 +08:00
fzyzcjy
8880a5cd5c
Update scheduler.py 2024-03-23 11:27:33 +08:00
Ross Wightman
c559c3911f Improve vit conversions. OpenAI convert pass through main convert for patch & pos resize. Fix #2120 2024-03-21 10:00:43 -07:00
Ross Wightman
256cf19148 Rename tinyclip models to fit existing 'clip' variants, use consistently mapped OpenCLIP compatible checkpoint on hf hub 2024-03-20 15:21:46 -07:00
Thien Tran
1a1d07d479 add other tinyclip 2024-03-19 07:27:09 +08:00
Thien Tran
dfffffac55 add tinyclip 8m 2024-03-19 07:02:17 +08:00
Ross Wightman
6ccb7d6a7c
Merge pull request #2111 from jamesljlster/enhance_vit_get_intermediate_layers
Vision Transformer (ViT) get_intermediate_layers: enhanced to support dynamic image size and saved computational costs from unused blocks
2024-03-18 13:41:18 -07:00
Cheng-Ling Lai
db06b56d34
Saved computational costs of get_intermediate_layers() from unused blocks 2024-03-17 21:34:06 +08:00
Cheng-Ling Lai
4731e4efc4
Modified ViT get_intermediate_layers() to support dynamic image size 2024-03-16 23:07:21 +08:00
Ross Wightman
ba641e07ae Add support for dynamo based onnx export 2024-03-13 12:05:26 -07:00
SmilingWolf
59cb0be595 SwinV2: add configurable act_layer argument
Defaults to "gelu", but makes it possible to pass "gelu_tanh".
Makes it easier to port weights from JAX/Flax, where the tanh
approximation is the default.
2024-03-05 22:04:17 +01:00
Ross Wightman
49992b0dc7
Update version.py
Update to 0.9.16 for release
2024-02-19 11:08:17 -08:00
Ross Wightman
35d6eef0df Version bump, add test markers back to toml 2024-02-16 09:04:00 -08:00
Ross Wightman
31e0dc0a5d Tweak hgnet before merge 2024-02-12 15:00:32 -08:00
Ross Wightman
3e03491e49 Merge branch 'master' of https://github.com/seefun/pytorch-image-models into seefun-master 2024-02-12 14:59:54 -08:00
Ross Wightman
958938845a
Update version.py 2024-02-10 23:10:50 -08:00
Ross Wightman
47c9bc4dc6 Fix device idx split 2024-02-10 21:41:14 -08:00
Ross Wightman
59239d9df5 Cleanup imports for vit relpos 2024-02-10 21:40:57 -08:00
Ross Wightman
ac1b08deb6 fix_init on vit & relpos vit 2024-02-10 20:15:37 -08:00
Ross Wightman
935950cc11 Fix F.sdpa attn drop prob 2024-02-10 20:14:47 -08:00
Ross Wightman
0737cf231d Add Next-ViT 2024-02-10 17:05:16 -08:00
Ross Wightman
d6c2cc91af Make NormMlpClassifier head reset args consistent with ClassifierHead 2024-02-10 16:25:33 -08:00
Ross Wightman
87fec3dc14 Update experimental vit model configs 2024-02-10 16:05:58 -08:00
Ross Wightman
7d3c2dc993 Add group_matcher for DaViT 2024-02-10 14:58:45 -08:00
Ross Wightman
7bc7798d0e Type annotation correctness for create_act 2024-02-10 14:57:58 -08:00
Ross Wightman
7d121ac2ef Small tweak of timm ToTensor for clarity 2024-02-10 14:57:40 -08:00
Ross Wightman
a08b57e801 Fix distributed flag bug w/ flex device handling 2024-02-03 16:26:15 -08:00
Ross Wightman
bee0471f91 forward() pass through for ema model, flag for ema warmup, comment about warmup 2024-02-03 16:24:45 -08:00
Ross Wightman
5e4a4b2adc Merge branch 'device_flex' into mesa_ema 2024-02-02 09:45:30 -08:00
Ross Wightman
dd84ef2cd5 ModelEmaV3 and MESA experiments 2024-02-02 09:45:04 -08:00
Ross Wightman
d0ff315eed Merge remote-tracking branch 'emav3/faster_ema' into mesa_ema 2024-01-27 14:52:10 -08:00
Ross Wightman
88889de923 Fix meshgrid deprecation warnings and backward compat with explicit 'ndgrid' and 'meshgrid' fn w/o indexing arg 2024-01-27 13:48:33 -08:00
Ross Wightman
d4386219c6 Improve type handling for arange & rel pos embeds, keep calculations in float32 until application (may change to apply in float32 in future). Prevent arange type hijacking by DeepSpeed Zero 2024-01-26 16:35:51 -08:00
Ross Wightman
3234daf783 Add missing deprecation mapping for a densenet and xcit model. Fix #2086. Tweak xcit pos embed use of arange for better low prec safety. 2024-01-24 22:04:04 -08:00
Ross Wightman
809a9e14e2 Pass train-crop-mode to create_loader/transforms from train.py args 2024-01-24 16:19:02 -08:00
Ross Wightman
a48ab818f5 Improving device flexibility in train. Fix #2081 2024-01-20 15:10:20 -08:00
Li zhuoqun
53a4888328 Add droppath and type hint to Xception. 2024-01-19 11:15:47 -08:00
kalazus
7f19a4cce7 fix fast catavgmax selection 2024-01-16 10:30:08 -08:00
Ross Wightman
2eac2f6955 Fiddling with iterator wrapping for HF ds streaming 2024-01-09 12:41:54 -08:00
Ross Wightman
992976f007
Update version.py 2024-01-08 09:39:22 -08:00
Ross Wightman
c50004db79 Allow training w/o validation split set 2024-01-08 09:38:42 -08:00
Ross Wightman
be0944edae Significant transforms, dataset, dataloading enhancements. 2024-01-08 09:38:42 -08:00
Ross Wightman
b5a4fa9c3b Add pos_weight and support for summing over classes to BCE impl in train scripts 2023-12-30 12:13:06 -08:00
方曦
9dbea3bef6 fix cls head in hgnet 2023-12-27 21:26:26 +08:00
SeeFun
56ae8b906d
fix reset head in hgnet 2023-12-27 20:11:29 +08:00
SeeFun
6862c9850a
fix backward in hgnet 2023-12-27 16:49:37 +08:00
SeeFun
6cd28bc5c2
Merge branch 'huggingface:main' into master 2023-12-27 16:43:37 +08:00
Ross Wightman
f2fdd97e9f Add parsable json results output for train.py, tweak --pretrained-path to force head adaptation 2023-12-22 11:18:25 -08:00
LR
e0079c92da
Update eva.py (#2058)
* Update eva.py

When argument class token = False, self.cls_token = None.

Prevents error from attempting trunc_normal_ on None:
AttributeError: 'NoneType' object has no attribute 'uniform_'

* Update eva.py

fix
2023-12-16 15:10:45 -08:00