dodox/dppo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Commit Graph

  • b259157a31 Launch Phase 2: Complete DPPO paper replication main ys1087@partner.kit.edu 2025-08-27 22:52:19 +0200
  • cb9846484f Update experiment plan with validation results and WandB URLs ys1087@partner.kit.edu 2025-08-27 22:14:10 +0200
  • bda37869e1 Add remaining validation test scripts and D3IL installer ys1087@partner.kit.edu 2025-08-27 21:06:44 +0200
  • 314a3f3c06 Add comprehensive dev test scripts and update experiment plan ys1087@partner.kit.edu 2025-08-27 21:02:55 +0200
  • 7e800c9a33 Complete MuJoCo fix and validate hopper fine-tuning ys1087@partner.kit.edu 2025-08-27 18:27:02 +0200
  • d739fa5e5e Add robomimic transport test and update experiment plan ys1087@partner.kit.edu 2025-08-27 16:21:06 +0200
  • 826b55a2d2 Integrate HoReKa Intel compiler fix for mujoco-py ys1087@partner.kit.edu 2025-08-27 16:09:13 +0200
  • 2404a34c36 Add MuJoCo compilation debugging and continue validation tests ys1087@partner.kit.edu 2025-08-27 15:32:29 +0200
  • 3cf999c32e Update documentation and simplify experiment tracking ys1087@partner.kit.edu 2025-08-27 15:25:43 +0200
  • 0424a080c1 feat: HoReKa cluster adaptation and validation ys1087@partner.kit.edu 2025-08-27 14:01:51 +0200
  • 93ac652def Start full hopper pre-training production run ys1087@partner.kit.edu 2025-08-27 12:31:42 +0200
  • a67f474fc0 Clarify pre-training vs fine-tuning phases and dev test purpose ys1087@partner.kit.edu 2025-08-27 12:29:31 +0200
  • 80339cad52 Update experiment plan with successful WandB run ys1087@partner.kit.edu 2025-08-27 12:28:16 +0200
  • 5a458aac67 Configure personal WandB entity and clean up docs ys1087@partner.kit.edu 2025-08-27 12:24:39 +0200
  • d43a9e2b3c Fix WandB configuration for proper logging ys1087@partner.kit.edu 2025-08-27 12:23:43 +0200
  • e8e7233d98 Fix WandB config issue and achieve working DPPO setup ys1087@partner.kit.edu 2025-08-27 12:19:38 +0200
  • 7fc9b17871 Clean up experiment tracking ys1087@partner.kit.edu 2025-08-27 12:08:31 +0200
  • 4adf67694a Fix Hydra config error in dev script ys1087@partner.kit.edu 2025-08-27 12:07:38 +0200
  • f88a5be4fe Add experiment tracking plan ys1087@partner.kit.edu 2025-08-27 12:06:34 +0200
  • add21c7019 Clarify that installation must run on GPU node ys1087@partner.kit.edu 2025-08-27 12:03:41 +0200
  • 2be39c4f2e Fix README: remove incorrect cluster policy reference ys1087@partner.kit.edu 2025-08-27 12:01:31 +0200
  • 835441af45 Fix broken image URL: use raw GitHub URL for cross-origin compatibility ys1087@partner.kit.edu 2025-08-27 12:00:59 +0200
  • 2bb63d0ed1 Fix README: remove local git configuration details ys1087@partner.kit.edu 2025-08-27 12:00:08 +0200
  • 30f59aaa9b Add HoReKa cluster documentation to README ys1087@partner.kit.edu 2025-08-27 11:59:32 +0200
  • 05dddfa10c Add HoReKa cluster setup and SLURM scripts ys1087@partner.kit.edu 2025-08-27 11:57:32 +0200
  • cc7234ad7f add note about ft_denoising_steps in eval in README allenzren 2025-02-04 11:50:53 -0500
  • b8086ed12e update version allenzren 2025-02-04 11:49:18 -0500
  • 9032d02eae change default ft_denoising_steps in eval configs to 0 (assume evaluating pre-trained models) allenzren 2025-02-04 11:48:59 -0500
  • fc42865c77 rename DiffusionEvalFT to DiffusionEval allenzren 2025-02-04 11:40:33 -0500
  • a746220905 allow loading pre-trained weights (not fine-tuned) in DiffusionEvalFT allenzren 2025-02-04 11:39:56 -0500
  • 169a16dda7 update eval configs allenzren 2025-01-03 17:22:38 -0500
  • ace2bbdab9 add separate eval model class that also initializes the pre-trained policy for early denoising steps allenzren 2025-01-03 17:22:27 -0500
  • e7f73dffc1 update batch size in D3IL so it works with the new form of gradient update allenzren 2024-12-24 02:06:17 -0500
  • 1d04211666 v0.7 (#26) Allen Z. Ren 2024-11-20 15:47:52 -0500
  • d2929f65e1
         update isaacgym download path Allen Z. Ren 2024-11-12 18:05:23 -0500
  • 7d1b3a236f update D3IL pre-processing, fix normalization bug in robomimic pre-processing allenzren 2024-11-08 18:40:42 -0500
  • 0bdae945e9 use default epoch_start_ema=20 and update_ema_freq=10 allenzren 2024-11-07 10:55:16 -0500
  • c0921a1fb5 remove update_ema_freq allenzren 2024-11-06 21:01:15 -0500
  • e1ef4ca1cf
         More frequent EMA update (#20) Allen Z. Ren 2024-11-06 20:42:31 -0500
  • dc8e0c9edc
         v0.6 (#18) Allen Z. Ren 2024-10-30 19:58:06 -0400
  • 7b10df690d fix diffusion loss when predicting initial noise allenzren 2024-10-13 11:19:10 -0400
  • 4e14b8086d fix itr initialization in eval agents allenzren 2024-10-10 14:10:16 -0400
  • e0842e71dc
         v0.5 to main (#10) Allen Z. Ren 2024-10-07 16:35:13 -0400
  • dd14c5887c set deterministic=True when sampling in diffusion evaluation allenzren 2024-09-26 01:15:10 -0400
  • 4962bbce38 rename train.py to run.py allenzren 2024-09-17 16:33:53 -0400
  • c9f24ba0c3 add evaluation agents and some example configs allenzren 2024-09-17 16:32:45 -0400
  • bc52beca1e add minor docs to diffusion classes and clean up some args allenzren 2024-09-17 16:26:25 -0400
  • ef5b14f820 fix observation history indexing in the dataset allenzren 2024-09-17 12:43:15 -0400
  • 1aaa6c2302 support varying img size allenzren 2024-09-16 17:55:31 -0400
  • 64595baca9 more intuitive handling of done in GAE allenzren 2024-09-13 16:29:56 -0400
  • f5a8da5719 typo allenzren 2024-09-11 21:50:02 -0400
  • f13eb203e1 allow history observation allenzren 2024-09-11 21:44:47 -0400
  • 2ddf63b8f5 squash commits allenzren 2024-09-11 21:09:17 -0400
  • 8ce0aa1485 simplify pre-training dataset, use npz allenzren 2024-09-08 17:52:16 -0400
  • 447c8dfd02 update instruction on reducing cpu threads allenzren 2024-09-08 13:43:52 -0400
  • a658353eb7 update compute instruction allenzren 2024-09-04 13:19:39 -0400
  • 771240b7a6 fix repo link allenzren 2024-09-03 21:05:26 -0400
  • 8293b0936b release allenzren 2024-09-03 21:03:27 -0400