Default Branch

a22366e3ce · Merge pull request #2503 from huggingface/beit3_remap_clean · Updated 2025-05-31 07:40:28 +08:00

Branches

57adc1acc8 · Fix rotary embed version of attn pool. Bit of cleanup/naming · Updated 2024-06-12 14:49:17 +08:00

451
0
Included

c63da1405c · Pretrained cfg name mismatch · Updated 2024-06-12 12:16:54 +08:00

451
0
Included

5ee06760dc · Fix classifier input dim for mnv3 after last changes · Updated 2024-06-08 04:53:13 +08:00

459
0
Included

7ccb10ebff · Disable efficient_builder debug flag · Updated 2024-06-07 12:50:27 +08:00

464
0
Included

5fa6efa158 · Add anti-aliasing support to mobilenetv3 and efficientnet family models. Update MobileNetV4 model defs, resolutions. Fix #599 · Updated 2024-05-28 13:06:22 +08:00

485
0
Included

286d941923 · Add teddy-bear class back to first 1000 classes of imagenet22k_ms_synsets (index 851) · Updated 2024-04-10 00:33:08 +08:00

573
0
Included

5c5ae8d401 · Fix #2132, remove use of _C.set_grad_enable. Line endings were messed up too · Updated 2024-04-10 00:00:23 +08:00

573
0
Included

17b892f703 · Fix #2139, disable strict weight loading when head changes from classification · Updated 2024-04-09 23:41:37 +08:00

573
0
Included

c559c3911f · Improve vit conversions. OpenAI convert pass through main convert for patch & pos resize. Fix #2120 · Updated 2024-03-22 01:00:43 +08:00

578
0
Included

ba641e07ae · Add support for dynamo based onnx export · Updated 2024-03-14 03:05:26 +08:00

590
0
Included

35d6eef0df · Version bump, add test markers back to toml · Updated 2024-02-17 01:04:00 +08:00

599
0
Included

47c9bc4dc6 · Fix device idx split · Updated 2024-02-11 13:41:14 +08:00

612
0
Included

9b25ded392 · Fix meshgrid deprecation warnings and backward compat with explicit 'ndgrid' and 'meshgrid' fn w/o indexing arg · Updated 2024-01-28 04:37:05 +08:00

633
1

284e4ea7a9 · 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 · Updated 2024-01-27 06:17:54 +08:00

638
1

f58e823a1c · Allow training w/o validation split set · Updated 2024-01-07 02:27:08 +08:00

649
2

db3c18999b · DFN CLIP ViT support · Updated 2023-11-01 01:22:27 +08:00

706
1

242611c5a1 · Added hub weights for dinov2 register models · Updated 2023-10-30 14:01:49 +08:00

715
2

e728f3efdb · Cleanup ijepa models, they're just gap (global-avg-pool) models w/o heads. fc-norm conversion was wrong, gigantic should have been giant · Updated 2023-10-18 06:44:46 +08:00

727
0
Included

7018abb00d · Another try · Updated 2023-10-13 00:37:59 +08:00

737
6

379780bb6c · Remove sdpa context mgrs · Updated 2023-09-26 14:30:56 +08:00

762
5