3bb5ed5e91
Extend reward classifier for multiple camera views ( #626 )
2025-01-13 13:57:49 +01:00
c5bca1cf0f
[Port HIL_SERL] Final fixes for the Reward Classifier ( #598 )
2025-01-06 11:34:00 +01:00
35de91ef2b
added temporary fix for missing task_index key in online environment
2024-12-30 13:47:28 +00:00
ee306e2f9b
split encoder for critic and actor
2024-12-29 23:59:39 +00:00
bae3b02928
style fixes
2024-12-29 14:35:21 +00:00
5b4adc00bb
Refactor SAC configuration and policy for improved action sampling and stability
...
- Updated SACConfig to replace standard deviation parameterization with log_std_min and log_std_max for better control over action distributions.
- Modified SACPolicy to streamline action selection and log probability calculations, enhancing stochastic behavior.
- Removed deprecated TanhMultivariateNormalDiag class to simplify the codebase and improve maintainability.
These changes aim to enhance the robustness and performance of the SAC implementation during training and inference.
2024-12-29 14:27:19 +00:00
22fbc9ea4a
Refine SAC configuration and policy for enhanced performance
...
- Updated standard deviation parameterization in SACConfig to 'softplus' with defined min and max values for improved stability.
- Modified action sampling in SACPolicy to use reparameterized sampling, ensuring better gradient flow and log probability calculations.
- Cleaned up log probability calculations in TanhMultivariateNormalDiag for clarity and efficiency.
- Increased evaluation frequency in YAML configuration to 50000 for more efficient training cycles.
These changes aim to enhance the robustness and performance of the SAC implementation during training and inference.
2024-12-29 14:21:49 +00:00
ca74a13d61
Refactor SACPolicy for improved action sampling and standard deviation handling
...
- Updated action selection to use distribution sampling and log probabilities for better stochastic behavior.
- Enhanced standard deviation clamping to prevent extreme values, ensuring stability in policy outputs.
- Cleaned up code by removing unnecessary comments and improving readability.
These changes aim to refine the SAC implementation, enhancing its robustness and performance during training and inference.
2024-12-29 14:17:25 +00:00
18a4598986
trying to get sac running
2024-12-29 14:14:13 +00:00
dc54d357ca
Added normalization schemes and style checks
2024-12-29 12:51:21 +00:00
08ec971086
added optimizer and sac to factory.py
2024-12-23 14:12:03 +01:00
b53d6e0ff2
[HIL-SERL PORT] Fix linter issues ( #588 )
2024-12-23 10:44:29 +01:00
70b652f791
[Port Hil-SERL] Add unit tests for the reward classifier & fix imports & check script ( #578 )
2024-12-23 10:43:55 +01:00
7b68bfb73b
added comments from kewang
2024-12-17 18:03:46 +01:00
7e0f20fbf2
Enhance SAC configuration and policy with new parameters and subsampling logic
...
- Added `num_subsample_critics`, `critic_target_update_weight`, and `utd_ratio` to SACConfig.
- Implemented target entropy calculation in SACPolicy if not provided.
- Introduced subsampling of critics to prevent overfitting during updates.
- Updated temperature loss calculation to use the new target entropy.
- Added comments for future UTD update implementation.
These changes improve the flexibility and performance of the SAC implementation.
2024-12-17 17:58:11 +01:00
def42ff487
Port SAC WIP ( #581 )
...
Co-authored-by: KeWang1017 <ke.wang@helloleap.ai>
2024-12-17 16:16:59 +01:00
c9af8e36a7
completed losses
2024-12-17 16:16:36 +01:00
ed66c92383
nit in control_robot.py
2024-12-17 11:04:56 +07:00
668d493bf9
Update lerobot/scripts/train_hilserl_classifier.py
...
Co-authored-by: Yoel <yoel.chornton@gmail.com>
2024-12-17 02:44:31 +07:00
67f4d7ea7a
LerobotDataset pushable to HF from any folder ( #563 )
2024-12-17 02:44:23 +07:00
4b0c88ff8e
Update 7_get_started_with_real_robot.md ( #559 )
2024-12-17 02:44:11 +07:00
b19fef9d18
Control simulated robot with real leader ( #514 )
...
Co-authored-by: Remi <remi.cadene@huggingface.co>
2024-12-17 02:44:03 +07:00
1612e00e63
Fix missing local_files_only in record/replay ( #540 )
...
Co-authored-by: Simon Alibert <alibert.sim@gmail.com>
2024-12-17 02:43:10 +07:00
c3bc136420
Refactor OpenX ( #505 )
2024-12-17 02:42:59 +07:00
1020bc3108
Fixup
2024-12-17 02:42:53 +07:00
7fcf638c0d
Add human intervention mechanism and eval_robot script to evaluate policy on the robot ( #541 )
...
Co-authored-by: Yoel <yoel.chornton@gmail.com>
2024-12-17 02:41:31 +07:00
e35546f58e
Reward classifier and training ( #528 )
...
Co-authored-by: Daniel Ritchie <daniel@brainwavecollective.ai>
Co-authored-by: resolver101757 <kelster101757@hotmail.com>
Co-authored-by: Jannik Grothusen <56967823+J4nn1K@users.noreply.github.com>
Co-authored-by: Remi <re.cadene@gmail.com>
Co-authored-by: Michel Aractingi <michel.aractingi@huggingface.co>
2024-12-17 02:41:29 +07:00
1aa8d4ac91
nit
2024-12-17 02:39:15 +07:00
66f8736598
fixing typo from 'teloperation' to 'teleoperation' ( #566 )
2024-12-11 05:57:52 -08:00
4c41f6fcc6
Fix example 6 ( #572 )
2024-12-11 10:32:18 +01:00
44f9b21e74
LerobotDataset pushable to HF from any folder ( #563 )
2024-12-09 11:32:25 +01:00
03f49ceaf0
Update 7_get_started_with_real_robot.md ( #559 )
2024-12-09 00:17:49 +01:00
8e7d6970ea
Control simulated robot with real leader ( #514 )
...
Co-authored-by: Remi <remi.cadene@huggingface.co>
2024-12-03 12:20:05 +01:00
286bca37cc
Fix missing local_files_only in record/replay ( #540 )
...
Co-authored-by: Simon Alibert <alibert.sim@gmail.com>
2024-12-03 10:53:21 +01:00
a2c181992a
Refactor OpenX ( #505 )
2024-12-03 00:51:55 +01:00
32eb0cec8f
Dataset v2.0 ( #461 )
...
Co-authored-by: Remi <remi.cadene@huggingface.co>
2024-11-29 19:04:00 +01:00
96c7052777
Rename deprecated argument (temporal_ensemble_momentum) ( #490 )
2024-11-25 21:05:13 +01:00
975c1c25c3
Add distinction between two unallowed cases in name check "eval_" ( #489 )
2024-11-22 19:19:57 +01:00
20f466768e
bug causes error uploading to huggingface, unicode issue on windows. ( #450 )
2024-11-22 19:15:58 +01:00
8af693548e
Add support for Windows ( #494 )
2024-11-22 19:14:25 +01:00
963738d983
fix: broken images and a few minor typos in README ( #499 )
...
Signed-off-by: ivelin <ivelin117@gmail.com>
2024-11-05 15:30:59 +01:00
e0df56de62
Fix config file ( #495 )
2024-10-31 16:41:49 +01:00
538455a965
feat: enable to use multiple rgb encoders per camera in diffusion policy ( #484 )
...
Co-authored-by: Alexander Soare <alexander.soare159@gmail.com>
2024-10-30 11:00:05 +01:00
172809a502
[Fix] Move back to manual calibration ( #488 )
2024-10-26 15:27:21 +02:00
55e4ff6742
Fix autocalib moss ( #486 )
2024-10-26 12:15:17 +02:00
07e8716315
Add FeetechMotorsBus, SO-100, Moss-v1 ( #419 )
...
Co-authored-by: jess-moss <jess.moss@huggingface.co>
Co-authored-by: Simon Alibert <75076266+aliberts@users.noreply.github.com>
2024-10-25 11:23:55 +02:00
114870d703
Fix link ( #482 )
...
Co-authored-by: Remi <remi.cadene@huggingface.co>
2024-10-23 16:24:06 +02:00
2efee45ef1
Update 9_use_aloha.md, missing comma ( #479 )
2024-10-23 16:13:26 +02:00
c351e1fff9
Fix gymnasium version as pre-1.0.0 ( #471 )
...
Co-authored-by: Remi <re.cadene@gmail.com>
Co-authored-by: Remi <remi.cadene@huggingface.co>
2024-10-18 10:23:27 +02:00
cd0fc261c0
Make `say(blocking=True)` work for Linux ( #460 )
2024-10-17 15:22:21 +01:00
Michel Aractingi
Eugene Mironov
KeWang1017
KeWang
Claudio Coppola
berjaoui
Remi
Yoel
s1lent4gnt
Simon Alibert
KasparSLT
Jannik Grothusen
resolver101757
Daniel Ritchie
Ivelin Ivanov
Arsen Ohanyan
Hirokazu Ishida
Bastian Krohg
Boris Zimka
Alexander Soare