Commit Graph

  • 051f6c6803 Merge branch 'main' into feat/decouple_record_script Steven Palma 2026-04-17 14:25:18 +02:00
  • c8df80ae91 Merge remote-tracking branch 'origin/main' into codex/model-profiling Pepijn 2026-04-17 12:27:11 +01:00
  • d4a229444b fix(ci): not fail when skipped (#3399) Steven Palma 2026-04-17 12:02:38 +02:00
  • 1ac8e96575 refactor(profiling): shrink lerobot_train.py diff via start()/finalize() Pepijn 2026-04-17 10:59:43 +01:00
  • 098ebb4d72 feat(ci): send slack notification if latest dependecy test is broken (#3398) Steven Palma 2026-04-17 11:28:24 +02:00
  • a6dd28e8b4 fix(profiling): tolerate groot dep-install failure Pepijn 2026-04-16 21:15:14 +02:00
  • 1842100402 feat(profiling): record forward/backward/optimizer timings Pepijn 2026-04-16 20:26:27 +02:00
  • 00e9defb80 fix(profiling): build flash-attn without isolation for groot Pepijn 2026-04-16 20:21:58 +02:00
  • b81eef43c8 fix(profiling): wall_x OOM and xvla rename_map Pepijn 2026-04-16 19:49:12 +02:00
  • d483dd4c4b feat(profiling): profile groot, xvla, diffusion, wall_x on PRs Pepijn 2026-04-16 19:01:31 +02:00
  • a56423fa33 Merge branch 'main' into codex/model-profiling Pepijn 2026-04-16 18:58:35 +02:00
  • 78201f3226 Merge branch 'main' into feat/robotwin-benchmark Pepijn 2026-04-16 18:57:39 +02:00
  • 4ccc4e9a66 fix(docs): use plain markdown image to fix MDX build Pepijn 2026-04-16 18:57:27 +02:00
  • fdbbc35cca fix(docs): use correct RoboTwin 2.0 teaser image URL Pepijn 2026-04-16 18:54:12 +02:00
  • 741a6d5246 fix(docs): correct RoboTwin 2.0 paper arxiv link Pepijn 2026-04-16 18:53:55 +02:00
  • a4102ee86d fix: integrate PR #3315 review feedback Pepijn 2026-04-16 18:33:53 +02:00
  • 9bc2df80bb chore(docs): adding a jupyter notebook that gives you ready-to-paste commands (#3395) Maxime Ellerbach 2026-04-16 17:53:35 +02:00
  • 04ae0312a2 HW tests fixes Steven Palma 2026-04-16 17:29:22 +02:00
  • cc634de9e7 add docstrings Steven Palma 2026-04-16 16:40:33 +02:00
  • da7da741f1 fix(profiling): use SGD for pi0/pi05/pi0_fast and free CUDA cache after deterministic forward Pepijn 2026-04-16 16:09:56 +02:00
  • b1e16783de refactor: extract profiling into self-contained TrainingProfiler class Pepijn 2026-04-16 16:00:49 +02:00
  • 3eda5712d3 some more iterations Steven Palma 2026-04-16 15:52:23 +02:00
  • d7e25c8326 refactor(rl): expose public API in rl/__init__ and use relative imports in sub-packages Khalil Meftah 2026-04-16 15:46:34 +02:00
  • a4544ffea7 fix(profiling): use bf16 dtype and gradient checkpointing for pi0/pi05 Pepijn 2026-04-16 15:35:25 +02:00
  • dbe01b0444 fix(profiling): fix pi0 cuBLAS error and pi05 OOM on 22GB GPU Pepijn 2026-04-16 15:34:04 +02:00
  • e16a95a78e refactor(profiling): remove cProfile, keep torch profiler only Pepijn 2026-04-16 15:32:59 +02:00
  • 4137b5785d fix(profiling): align libero smoke specs with pretrained policies Pepijn 2026-04-16 15:11:54 +02:00
  • 8ece10e484 feat(ci): profile more models in pr smoke runs Pepijn 2026-04-16 14:49:37 +02:00
  • ddeb216ab9 fix(ci): skip hub publish for pr profiling runs Pepijn 2026-04-16 14:37:17 +02:00
  • 783ec6e232 minor improvements Steven Palma 2026-04-16 14:34:22 +02:00
  • d46d67f75d fix(profiling): forward GIT_REF + PR_NUMBER into Docker container Pepijn 2026-04-16 13:38:13 +02:00
  • b746cd3c61 fix(profiling): sort import + move expressions to env vars for zizmor Pepijn 2026-04-16 13:30:13 +02:00
  • 6d1a5fca02 fix(profiling): keep ci green when hub publish is unauthorized Pepijn 2026-04-16 13:07:30 +02:00
  • 8d7099cd7d fix(profiling): publish preview runs via hf dataset prs Pepijn 2026-04-16 12:50:57 +02:00
  • 516f39685a fix(profiling): skip dataset creation on publish Pepijn 2026-04-16 12:09:03 +02:00
  • b27e838376 fix(profiling): publish preview rows to existing dataset Pepijn 2026-04-16 11:54:35 +02:00
  • 40470648d1 feat(profiling): publish preview runs for dashboard debugging Pepijn 2026-04-16 10:54:34 +02:00
  • 25e5062b2c fix(profiling): read generic device timings from profiler Pepijn 2026-04-16 10:29:01 +02:00
  • 35e3b28da1 fix(profiling): normalize timing metrics before export Pepijn 2026-04-16 10:11:14 +02:00
  • ed8a98dda6 fix(profiling): preserve policy mode for deterministic forward Pepijn 2026-04-16 09:50:29 +02:00
  • 9dc38d9993 fix(ci): isolate torch cache in profiling job Pepijn 2026-04-16 09:32:16 +02:00
  • 3922f81791 fix(ci): set HF_LEROBOT_HOME in profiling job Pepijn 2026-04-15 23:35:27 +02:00
  • 28e8483297 fix(ci): disable policy hub push in profiling runs Pepijn 2026-04-15 23:02:28 +02:00
  • e1b22ed1c4 fix(ci): set torchinductor cache dir in profiling job Pepijn 2026-04-15 22:55:31 +02:00
  • f2d0f04dd0 fix(ci): isolate profiling container home dirs Pepijn 2026-04-15 22:51:22 +02:00
  • 3ea722c6c0 fix(ci): run profiling container as runner user Pepijn 2026-04-15 22:47:29 +02:00
  • 48660e7a7c fix(ci): avoid host shell expansion in policy error Pepijn 2026-04-15 22:42:34 +02:00
  • c94fe868c9 fix(ci): install only profiling policy extras Pepijn 2026-04-15 22:38:37 +02:00
  • d4f27cfb6e fix(ci): restore docker env line continuation Pepijn 2026-04-15 22:33:14 +02:00
  • 1a2aec1b04 feat(profiling): add weekly model profiling Pepijn 2026-04-15 22:31:44 +02:00
  • 2ab59a3099 feat(benchmarks): add matrix runner and leaderboard feat/libero-benchmark Pepijn 2026-04-15 21:31:33 +02:00
  • 4e3175ff15 address review Steven Palma 2026-04-15 19:31:53 +02:00
  • edd7fc52a8 feat: introduce inference engine strategy Steven Palma 2026-04-15 17:51:44 +02:00
  • a5ad273b62 fix(tests): skip tests that require grpc if not available Khalil Meftah 2026-04-15 16:30:20 +02:00
  • 0f0f8b8961 imports and comments Steven Palma 2026-04-15 16:28:56 +02:00
  • 23bece96a4 fix(tests): ensure tensor stats comparison accounts for reshaping in normalization tests Khalil Meftah 2026-04-15 16:12:08 +02:00
  • 7a1c9e74c3 fix: skip tests that require grpc if not available Khalil Meftah 2026-04-15 15:18:04 +02:00
  • bd74f6733d chore: bump doc-builder SHA for PR upload workflow (#3386) Remy 2026-04-15 12:15:24 +02:00
  • c88cf979f1 fix: use string key for IS_INTERVENTION in complementary_info to avoid torch.load serialization error Khalil Meftah 2026-04-15 11:49:38 +02:00
  • ad9662b4a8 fix(robotwin): defer YAML lookup and realign tests with current API Pepijn 2026-04-15 11:14:02 +02:00
  • 79db54dc34 Merge branch 'main' into feat/decouple_record_script Steven Palma 2026-04-15 11:06:45 +02:00
  • f291d3bfa9 docs(robotwin): add robotwin to _toctree.yml under Benchmarks Pepijn 2026-04-15 11:06:34 +02:00
  • 49186359b0 refactor(robotwin): rebase docker image on huggingface/lerobot-gpu Pepijn 2026-04-15 11:05:07 +02:00
  • 6f4a96333e chore(docs): update contributing (#3387) Steven Palma 2026-04-15 11:02:37 +02:00
  • 99792bb17b ci: point benchmark eval checkpoints at the lerobot/ org mirrors Pepijn 2026-04-15 11:02:06 +02:00
  • 6ae07878f7 Merge branch 'main' into feat/decouple_record_script Steven Palma 2026-04-14 22:54:29 +02:00
  • 9021d2d240 refactor(imports): enforce guard pattern (#3382) Steven Palma 2026-04-14 22:54:05 +02:00
  • 10d05e03bc Merge branch 'main' into feat/decouple_record_script Steven Palma 2026-04-14 21:35:26 +02:00
  • e67ceb213d feat(robotwin): eval 5 diverse tasks per CI run with NL descriptions Pepijn 2026-04-14 21:03:15 +02:00
  • 793f52e360 fix(robotwin): install av-dep so lerobot_eval can write rollout MP4s Pepijn 2026-04-14 20:24:03 +02:00
  • ae113e0d99 fix(robotwin): expose _max_episode_steps for lerobot_eval.rollout Pepijn 2026-04-14 20:00:36 +02:00
  • 61a0269560 fix(robotwin): align observation_space dims with D435 camera output Pepijn 2026-04-14 19:35:01 +02:00
  • c2160ca86e refactor(robotwin): drop defensive dict guards, cache black fallback frame Pepijn 2026-04-14 18:39:40 +02:00
  • f40b30202b fix(robotwin): read nested get_obs() output and use aloha-agilex camera names Pepijn 2026-04-14 18:25:08 +02:00
  • 60e7d67cb8 fix: catch KeyboardInterrupt in safe_stop_image_writer to prevent corrupted frames (#3381) Khalil Meftah 2026-04-14 18:22:56 +02:00
  • 79a9ebdaa6 fix: add try/finally to control_loop to ensure image writer cleanup on exit Khalil Meftah 2026-04-14 17:54:35 +02:00
  • f2c29d78cf more improvements and fixes Steven Palma 2026-04-14 17:51:03 +02:00
  • b06f134fe4 fix(robotwin): re-enable autograd for CuRobo planner warmup and take_action Pepijn 2026-04-14 17:39:21 +02:00
  • da6e36fd03 Merge remote-tracking branch 'origin/main' into user/khalil-meftah/2026-02-16-rl-stack-refactor Khalil Meftah 2026-04-14 17:14:56 +02:00
  • 8bc47e4318 target review Steven Palma 2026-04-14 17:14:09 +02:00
  • 5558ea2207 feat(envs): add RoboTwin 2.0 benchmark integration Pepijn 2026-04-14 11:49:15 +02:00
  • 64dc08cb7b fix: include IS_INTERVENTION in complementary_info sent to learner for offline replay buffer Khalil Meftah 2026-04-14 16:35:08 +02:00
  • 49f32b9796 some more iterations Steven Palma 2026-04-14 16:34:52 +02:00
  • 1ede000bdd fix(rl): swap dict merge order to preserve teleop intervention flag (#3273) Radu 2026-04-14 07:20:54 -07:00
  • f55782f9f7 pre-commit run Steven Palma 2026-04-14 15:42:19 +02:00
  • 05a2604d6e first iteration Steven Palma 2026-04-14 15:42:04 +02:00
  • d57c58a532 fix: add thread synchronization to ReplayBuffer to prevent race condition between add() and sample() (#3372) Khalil Meftah 2026-04-14 13:16:45 +02:00
  • b3e76a92f2 fix(groot): compatibility fixes for gr00t in v0.5 (#3182) Matteo Tiezzi 2026-04-14 13:09:18 +02:00
  • f5c801fd34 fix(test): add missing device placement in multi-task DiT tests (#3349) Khalil Meftah 2026-04-14 12:25:29 +02:00
  • cff4bcf4a0 Update reward classifier training config (#3147) Ethan Pronovost 2026-04-14 02:28:49 -07:00
  • e6d282108d Fix: add kwargs in reward classifier __init__() Khalil Meftah 2026-04-14 11:13:43 +02:00
  • dab511dbb1 Merge branch 'main' into feat/libero-benchmark Pepijn 2026-04-14 10:43:49 +02:00
  • a656a982af fix(feetech): motor position readings overflow (#3373) Maxime Ellerbach 2026-04-13 22:39:58 +02:00
  • 187b2167ed feat(ci): benchmark smoke tests with isolated Docker images (LIBERO + MetaWorld) (#3319) Pepijn 2026-04-13 21:24:01 +02:00
  • a8838c081b perf: remove redundant CPU→GPU→CPU transition move in learner Khalil Meftah 2026-04-13 19:06:28 +02:00
  • ee0814ef60 refactor: update SACAlgorithm to pass action_dim to _init_critics and fix encoder reference Khalil Meftah 2026-04-13 18:31:17 +02:00
  • 7b0bdf2a98 fix: add thread synchronization to ReplayBuffer to prevent race condition between add() and sample() Khalil Meftah 2026-04-13 18:27:24 +02:00
  • 9bd844a3b9 fix(rl): ensure queue and process cleanup on abnormal exit (#3063) Jash Shah 2026-04-13 07:25:42 -07:00
  • 22d13d5859 reverting changes to pre-commit config file docs/feetech-wrist Maxime 2026-04-13 13:42:49 +02:00
  • a30526d8f9 trailing whitespace Maxime 2026-04-13 13:40:37 +02:00