All pull requests

Pull requests

• fix(datasets): reject non-finite delta timestamps before rounding

  #5114·Ramcharan49 opened Oct 27, 2026·

  ·

• fix(datasets): isolate nested tool schemas returned by metadata

  #5113·Ramcharan49 opened Oct 27, 2026·

  ·

• perf(groot): keep N1.7 camera frames on device during preprocessing

  policiestests
  #5112·Priyakarna13 opened Oct 27, 2026·3 of 5 tasks·

  ·
  1

• fix(smolvla): use fp32 trainable weights by default

  documentationpoliciestests
  #5111·jleem99 opened Oct 26, 2026·3 of 6 tasks·

  ·
  2

• fix(robots): expose B601 velocity and torque observations

  robotstests
  #5110·YidaWeng opened Oct 26, 2026·

  ·

• Feat/sim foundation

  #5109·s1lent4gnt opened Oct 26, 2026Member·

  ·

• fix(pi05): mask padded actions in the loss

  policiestests
  #5108·johyeongseob opened Oct 26, 2026·4 of 5 tasks·

  ·
  1

• fix(pi0): mask padded actions in the loss

  policiestests
  #5106·johyeongseob opened Oct 26, 2026·4 of 5 tasks·

  ·
  1

  1

• feat(optim): add the warmup_stable_decay scheduler

  #5103·HaomingSong opened Oct 26, 2026Collaborator·5/5·

  ·
  1

• chore(ci): run only the CUDA tests on the GPU job and build the large test models on the GPU

  #5101·Maximellerbach opened Oct 26, 2026Member·3/3·4 of 5 tasks·

  ·

• feat(record): record microphones in lerobot-record

  #5100·AlejandroParedesLT opened Oct 26, 2026·6 tasks·

  ·
  1

• feat(vla_jepa): opt-in world_model_rope_legacy for checkpoints trained before #4835

  documentationpoliciestests
  #5098·ravediamond opened Oct 26, 2026Collaborator·3 of 5 tasks·

  ·
  1

• fix(processor): keep pi05, pi0_fast and delta-action step settings across save/load

  #5097·ravediamond opened Oct 26, 2026Collaborator·2 of 5 tasks·

  ·
  1

• feat(train): configurable TF32 for training and evaluation

  configurationdocumentationtests
  #5096·HaomingSong opened Oct 26, 2026Collaborator·4/5·

  ·
  3

• feat(train): add FSDP2 MixedPrecisionPolicy and sync_each_batch

  #5095·HaomingSong opened Oct 26, 2026Collaborator·3/5·

  ·
  1

• test(inference): cover reset races at the rollout dispatch boundary

  #5093·theo-ai-lab opened Oct 26, 2026·2 of 5 tasks·

  ·
  2

• fix(eval): collect predicted latents through PEFT wrapped model

  #5092·HaomingSong opened Oct 25, 2026Collaborator·

  ·
  3

• fix(scripts): warn when fine-tuning replaces pretrained processor stats with dataset stats

  tests
  #5091·akushonkamen opened Oct 25, 2026·2 of 5 tasks·

  ·
  1

  2

• docs(openarm): nest recording flags under --dataset

  documentation
  #5090·ZainnQureshii opened Oct 25, 2026·2 of 5 tasks·

  ·
  2

• docs(eo1): update chunk_size and n_action_steps defaults to 16

  documentation
  #5089·ZainnQureshii opened Oct 25, 2026·2 of 5 tasks·

  ·
  2

• Fix SO-ARM101 hardware safety gate integration

  robotstests
  #5088·vantek988 opened Oct 25, 2026·

  ·
  3

• feat(train): support PEFT under sharded training (FSDP2, HSDP)

  configurationdocumentationpoliciestests
  #5086·HaomingSong opened Oct 25, 2026Collaborator·2/5·

  ·
  1

• fix(processor): preserve squeeze_batch_dim across save/load

  processortests
  #5085·not-heavychevy opened Oct 24, 2026·3 of 5 tasks·

  ·

• fix(xvla): apply the LIBERO env processors to LIBERO only and check the action conversion

  evaluationpolicies
  #5084·HaomingSong opened Oct 24, 2026Collaborator·

  ·
  2

• Add SO-ARM101 hardware safety gate

  robotstests
  #5083·vantek988 opened Oct 24, 2026·

  ·