mirrors
/
dinov2
mirror of https://github.com/facebookresearch/dinov2.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Commit Graph

  • 937a73bae7
         Merge 582883a633 into e1277af2ba Riadh Fezzani 2025-03-31 15:31:53 +0200
  • 582883a633 Replace xformers import warnings.warn by Logger.warning Riadh 2025-03-31 14:56:39 +0200
  • 3d03635c63 new dataset classes Veronikkkka 2025-03-26 12:51:56 +0000
  • 83f2d7ff62 Refactor training loop to incorporate effective epoch length and gradient accumulation steps ChuaHanChong 2025-03-25 06:06:52 +0000
  • 9bdb158362 Add scaling factor for gradient accumulation in forward_backward method ChuaHanChong 2025-03-25 06:06:38 +0000
  • 235eac76c9 Fix learning rate scaling to include gradient accumulation steps ChuaHanChong 2025-03-25 06:05:59 +0000
  • 3784ec00ce knn validation Veronikkkka 2025-03-21 19:06:32 +0000
  • 20db539127 a Veronikkkka 2025-03-20 19:04:32 +0000
  • 753b990c64 fsdp for raw adapter Veronikkkka 2025-03-18 10:22:05 +0000
  • 95b61eb500
         Add new ImageShipID datasets: 1M, 200k, and 25k (#7) Han Chong 2025-03-11 21:04:10 +0800
  • 63dfb318ec standart Veronikkkka 2025-03-11 11:47:21 +0000
  • 7d5c09e8e4 all changes commented Veronikkkka 2025-03-11 07:44:59 +0000
  • 693b581c24 added tensorboard Veronikkkka 2025-03-10 17:43:16 +0000
  • e68f8e290a add configuration for merge block indexes Veronikkkka 2025-03-10 17:13:41 +0000
  • 050e5677c8 interpolation before patch embedding Veronikkkka 2025-03-10 10:15:55 +0000
  • badc5ddfc8 patch embedding for ada Veronikkkka 2025-03-09 12:25:56 +0000
  • 8cb907ab93 init layer linear Veronikkkka 2025-03-05 12:55:26 +0000
  • 9c1bde505b added input, model, merge block for raw Veronikkkka 2025-03-05 09:00:59 +0000
  • 0909dbc557
         ViT-H/14 variant (#6) Han Chong 2025-02-27 17:54:50 +0800
  • 391342d78d
         New data (#5) Han Chong 2025-02-27 17:51:41 +0800
  • fd9fb1f8d3 make dinov2 support export to onnx Wala-Touati 2025-02-06 20:55:21 +0000
  • c9406038a2
         Add BTD and LAL (#4) Han Chong 2025-02-06 09:39:35 +0800
  • d00d2f5822
         Rename dataset (#3) Han Chong 2025-01-30 21:26:44 +0800
  • 7f08b158a1 full pipeline of scalekd arda 2025-01-13 16:38:17 +0000
  • 93aad6083b scalekd arda 2025-01-11 15:39:09 +0000
  • 180ed7bbd3
         Merge 065764ad21 into e1277af2ba dmitrysarov 2024-12-19 09:03:05 +0100
  • 065764ad21 mem efficient cosine scheduler Dmitrii Victorovich Shaulskii 2024-12-19 09:00:04 +0100
  • 69b50c8ff2 Added code for continuing training. arda 2024-12-01 04:47:56 +0000
  • 26d9ca6d12 Remove distillation folder arda 2024-12-01 03:32:39 +0000
  • c38640be55 Remove distillation folder arda 2024-12-01 03:30:00 +0000
  • 33172fba61 distillation arda 2024-12-01 03:28:31 +0000
  • 94aa2280b4 train nextvit on one node one gpu! liyubrt 2024-11-30 13:46:09 -0800
  • 36d54aaad2 debug nextvit training (tbc) liyubrt 2024-11-30 13:03:57 -0800
  • d683211ae3 work for 1 gpu vit model liyubrt 2024-11-30 08:07:03 -0800
  • 2aae191d0b add init code for distill liyubrt 2024-11-30 07:49:29 -0800
  • d84861dffb Test Chen Marvin 2024-11-23 07:22:09 +0100
  • 9eac5f063e Add onnx conversion script example, install onnx Wala-Touati 2024-11-18 16:56:20 +0000
  • 235f50669e Parametrizable attn_class in models/ViT Wala-Touati 2024-11-18 16:40:00 +0000
  • ace39d1b53 Change attention type in backbones hubconf Wala-Touati 2024-11-18 16:26:23 +0000
  • ca6209c24c
         Merge 85a2460209 into e1277af2ba Federico Baldassarre 2024-10-26 02:42:14 +0200
  • 85a2460209
         Fix: x.item() can not be called when creating the model on a "meta" device Federico Baldassarre 2024-10-26 02:19:00 +0200
  • d06d233ba1
         Merge 3a3217b0c0 into e1277af2ba Federico Baldassarre 2024-10-26 02:13:16 +0200
  • 3a3217b0c0
         Defer LayerScale initialization for compatibility with "meta" devices Federico Baldassarre 2024-10-26 01:30:36 +0200
  • 400983cf83 make training on custom data work! liyubrt 2024-10-22 19:51:52 -0700
  • 5917521c34
         Add ViT-H (#2) Han Chong 2024-10-03 15:06:50 +0800
  • 18ef3c8f7e
         Get attention outputs (#1) Han Chong 2024-10-01 13:54:36 +0800
  • 2eff634ba0 Add custom dataset ChuaHanChong 2024-10-01 05:48:31 +0000
  • 6c9ebd7311 ensure data locality in our infinite distributed sampler Adrian Wolny 2024-08-28 15:40:20 +0200
  • cd75745578 make sure training works with torchrun on the slurm cluster Adrian Wolny 2024-08-26 16:59:30 +0200
  • 2039cf4dfd add s3 support to the histo dataset Adrian Wolny 2024-08-26 12:08:55 +0200
  • 90b28006b1 integrate histo dataset and sampler Adrian Wolny 2024-08-23 00:40:12 +0200
  • 9164a24a56 setup wandb logging Adrian Wolny 2024-08-21 15:54:37 +0200
  • 5d31bf88fc make FSDP compatible with PyTorch 2.4 Adrian Wolny 2024-08-20 18:06:13 +0200
  • fb262e9d94
         Merge 1f3073a94f into e1277af2ba Setepenre 2024-08-07 09:44:54 -0400
  • 1f3073a94f - pierre.delaunay 2024-08-07 09:44:20 -0400
  • bbe11eba90
         Merge ac07cc88b6 into e1277af2ba Karim Abou Zeid 2024-07-20 15:19:40 +0200
  • ac07cc88b6
         Remove unnecessary warnings Karim Abou Zeid 2024-07-20 15:18:15 +0200
  • 52c0aec7fb
         Merge d283fc2b96 into e1277af2ba Sanskar Agrawal 2024-07-12 12:51:21 +0000
  • d283fc2b96 Improve readability Sanskar Agrawal 2024-07-12 18:16:33 +0530
  • b82eafd36b
         Merge a392ec9b2c into e1277af2ba eightsuzuki 2024-06-27 08:35:43 +0000
  • a392ec9b2c Update PyTorch extra-index-url to CUDA 11.8 eightsuzuki 2024-06-27 17:32:25 +0900
  • 90899607fa
         Merge 04b91003ba into e1277af2ba Sahil Mehra 2024-05-30 06:50:48 +0000
  • 04b91003ba
         Update README.md Sahil Mehra 2024-05-29 23:47:24 -0700
  • e4f4381337 add workflow Etienne Guevel 2024-05-29 17:33:35 +0200
  • d2f9b593d4 edit tests + README Etienne Guevel 2024-05-29 17:17:33 +0200
  • be41e7b2d3 Add env for cpu Etienne Guevel 2024-05-29 13:50:56 +0200
  • dc63c4c5f0 Merge branch 'RGB' of github.com:etienneguevel/dinov2 into RGB Etienne Guevel 2024-05-26 01:07:55 +0200
  • d605692003 fix stuff Etienne Guevel 2024-05-26 01:00:55 +0200
  • 279ec583f9 fix stuff Etienne Guevel 2024-05-26 01:00:55 +0200
  • 9d6d6d42b5 Merge branch 'RGB' of github.com:etienneguevel/dinov2 into RGB Etienne Guevel 2024-05-24 17:52:19 +0200
  • 0a0b3c13ac train changes for data preservation Etienne Guevel 2024-05-24 17:46:07 +0200
  • 996dd33131 train changes for data preservation Etienne Guevel 2024-05-24 17:46:07 +0200
  • 8da3dfab58 Merge branch 'RGB' of github.com:etienneguevel/dinov2 into RGB Etienne Guevel 2024-05-24 17:16:54 +0200
  • 619d4edb6d Merge branch 'RGB' of github.com:etienneguevel/dinov2 into RGB Etienne Guevel 2024-05-24 17:13:05 +0200
  • 84dd00b098 Merge branch 'RGB' of github.com:etienneguevel/dinov2 into RGB Etienne Guevel 2024-05-24 17:13:05 +0200
  • 9ef866789f preserve lab data changes Etienne Guevel 2024-05-24 16:55:10 +0200
  • 2e05e6b4ef preserve lab data changes Etienne Guevel 2024-05-24 16:55:10 +0200
  • e73c100c44
         Merge 479acef8f1 into e1277af2ba JW 2024-05-24 10:10:08 +0000
  • 479acef8f1 add the sys.path way to help finding the dinov2 in using VSCODE with linux_x64 JW 2024-05-24 18:05:27 +0800
  • 6684955cef fix RGB issue Etienne Guevel 2024-05-16 23:02:51 +0200
  • 6aaa74ce49 convert RGB Etienne Guevel 2024-05-16 12:10:38 +0200
  • 9e43e49463 modifs for RGBA Etienne Guevel 2024-05-16 11:46:01 +0200
  • 252ca3b3ab fix mem_per_gpu Etienne Guevel 2024-05-15 16:52:13 +0200
  • 874ac5192e add mem-per-gpu Etienne Guevel 2024-05-15 16:17:15 +0200
  • 087a4cea01 _handle and _streams fix Etienne Guevel 2024-05-15 16:01:34 +0200
  • 0e1c5b4e35 _handles change Etienne Guevel 2024-05-15 10:18:33 +0200
  • 12e4811549 initial changes Etienne Guevel 2024-05-14 15:11:51 +0200
  • 1054f48e75
         Merge 5a48d47656 into e1277af2ba Sepehr Sameni 2024-04-27 08:50:40 +0000
  • 5a48d47656
         Update transforms.py Sepehr Sameni 2024-04-27 10:31:28 +0200
  • 67b457e1b3
         Merge b8d0c3f391 into e1277af2ba Lance Legel 2024-04-19 14:40:45 +0000
  • b8d0c3f391 multi-class classification with DinoV2 Lance Legel 2024-04-19 10:40:36 -0400
  • b05584f6f8
         Add Pytorch 2.1 compatibility https://github.com/facebookresearch/dinov2/pull/281 Tuatini Godard 2024-04-17 14:31:30 +0200
  • 1914862536
         Merge 4547e69f19 into e1277af2ba D. Kobus 2024-03-29 17:18:58 +0000
  • 4547e69f19 fix names and cuda label D. Kobus 2024-03-29 18:11:38 +0100
  • 4b57d533bf fix conda env D. Kobus 2024-03-29 13:19:19 +0100
  • 464cb52bf0 bug fix in launch script cm090999 2024-03-14 10:21:11 +0100
  • 91edd274e6 required changes in config cm090999 2024-03-14 10:20:33 +0100
  • ee1dff98c6 add drop_path2 cm090999 2024-03-14 10:19:23 +0100
  • 602c340026 add original model generation cm090999 2024-03-14 10:19:10 +0100
  • e9bbd22734 removed unnnecessairy bug fix cm090999 2024-03-14 10:18:25 +0100