Merge branch 'main' into feat/language-annotation-pipeline

Resolves conflicts from 32 commits on main:

* docs/source/_toctree.yml — keep both new toc entries
  (annotation_pipeline + video_encoding_parameters).
* docs/source/language_and_recipes.mdx — adopt main's section
  ordering (Layer 2 before "Temporal semantics") and float32
  timestamp dtype to match the codebase.
* src/lerobot/configs/__init__.py — keep both export sets
  (recipe + video encoder).
* src/lerobot/datasets/dataset_metadata.py — drop redundant lazy
  imports (top-level imports cover both LANGUAGE_COLUMNS and
  DEFAULT_TOOLS); adopt main's @tools.setter for info.json
  write-back.
* src/lerobot/datasets/feature_utils.py — call the real
  validate_feature_language() instead of returning "".
* src/lerobot/datasets/language.py — float32 timestamps to match
  pa.float32() used in video_utils.py and the rest of the codebase.
* src/lerobot/datasets/language_render.py — adopt main's
  unwrap_scalar() helper (drops two hand-rolled .item()/list
  unwrappers); float32 in docstring.
* src/lerobot/processor/render_messages_processor.py — drop
  PR-local _scalar() helper, use shared unwrap_scalar().
* tests/datasets/test_language.py — adopt main's new float32 dtype
  + validate_feature_language warning tests.
* tests/datasets/test_dataset_metadata.py — adopt main's new
  tools.setter persist/clear tests.
* uv.lock — regenerated cleanly from main's resolver.

90 of 92 touched tests pass. Two pre-existing test failures
(test_module1_plan_memory_subtask_smoke,
test_module2_mid_episode_emits_paired_interjection_and_speech in
tests/annotations/test_modules.py) are unrelated to this merge —
that test file doesn't exist on main, so the failures originate on
the branch and are addressed by the 8 newer fix(annotate) commits
already on origin that will land in a follow-up.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Pepijn
2026-05-25 15:46:32 +02:00
182 changed files with 13664 additions and 5019 deletions
+38 -5
View File
@@ -59,8 +59,8 @@ keywords = ["lerobot", "huggingface", "robotics", "machine learning", "artifici
dependencies = [
# Core ML
"torch>=2.7,<2.11.0",
"torchvision>=0.22.0,<0.26.0",
"torch>=2.7,<2.12.0",
"torchvision>=0.22.0,<0.27.0",
"numpy>=2.0.0,<2.3.0", # NOTE: Explicitly listing numpy helps the resolver converge faster. Upper bound imposed by opencv-python-headless.
"opencv-python-headless>=4.9.0,<4.14.0",
"Pillow>=10.0.0,<13.0.0",
@@ -99,7 +99,18 @@ dataset = [
"pandas>=2.0.0,<3.0.0", # NOTE: Transitive dependency of datasets
"pyarrow>=21.0.0,<30.0.0", # NOTE: Transitive dependency of datasets
"lerobot[av-dep]",
"torchcodec>=0.3.0,<0.11.0; sys_platform != 'win32' and (sys_platform != 'linux' or (platform_machine != 'aarch64' and platform_machine != 'arm64' and platform_machine != 'armv7l')) and (sys_platform != 'darwin' or platform_machine != 'x86_64')", # NOTE: Windows support starts at version 0.7 (needs torch==2.8), ffmpeg>=8 support starts at version 0.8.1 (needs torch==2.9), system-wide ffmpeg support starts at version 0.10 (needs torch==2.10).
# NOTE: torchcodec wheel availability matrix (PyPI):
# - linux x86_64/amd64 + macOS arm64 : wheels since 0.3.0 (the historic supported set).
# - win32 x86_64 : wheels since 0.7.0 (needs torch>=2.8).
# - linux aarch64/arm64 : wheels since 0.11.0 (needs torch>=2.11).
# - macOS x86_64 (Intel) and linux armv7l: no wheels in any released version -> fall through to the PyAV decoder.
# Each platform gets its own line so the resolver picks the minimum version that has a wheel for it.
# Other torch/torchcodec pairings (informational): 0.8.1 = ffmpeg>=8 support, 0.10 = system-wide ffmpeg support, 0.12 needs torch==2.12.
"torchcodec>=0.3.0,<0.12.0; (sys_platform == 'linux' and (platform_machine == 'x86_64' or platform_machine == 'AMD64')) or (sys_platform == 'darwin' and platform_machine == 'arm64')",
"torchcodec>=0.7.0,<0.12.0; sys_platform == 'win32'",
"torchcodec>=0.11.0,<0.12.0; sys_platform == 'linux' and (platform_machine == 'aarch64' or platform_machine == 'arm64')",
"jsonlines>=4.0.0,<5.0.0",
]
training = [
@@ -127,7 +138,9 @@ dataset_viz = ["lerobot[dataset]", "lerobot[viz]"]
# Common
av-dep = ["av>=15.0.0,<16.0.0"]
pygame-dep = ["pygame>=2.5.1,<2.7.0"]
placo-dep = ["placo>=0.9.6,<0.9.17"]
# NOTE: 0.9.16 links against liburdfdom_sensor.so.4, which is unavailable on Ubuntu 24.04
# (noble ships urdfdom 3.x). Cap below 0.9.16 until system urdfdom 4.x is broadly available.
placo-dep = ["placo>=0.9.6,<0.9.16"]
transformers-dep = ["transformers>=5.4.0,<5.6.0"]
grpcio-dep = ["grpcio==1.73.1", "protobuf>=6.31.1,<6.32.0"]
can-dep = ["python-can>=4.2.0,<5.0.0"]
@@ -140,6 +153,8 @@ pyserial-dep = ["pyserial>=3.5,<4.0"]
deepdiff-dep = ["deepdiff>=7.0.1,<9.0.0"]
pynput-dep = ["pynput>=1.7.8,<1.9.0"]
pyzmq-dep = ["pyzmq>=26.2.1,<28.0.0"]
motorbridge-dep = ["motorbridge>=0.3.2,<0.4.0"]
motorbridge-smart-servo-dep = ["motorbridge-smart-servo>=0.0.4,<0.1.0"]
# Motors
feetech = ["feetech-servo-sdk>=1.0.0,<2.0.0", "lerobot[pyserial-dep]", "lerobot[deepdiff-dep]"]
@@ -163,6 +178,9 @@ unitree_g1 = [
"lerobot[pygame-dep]",
]
reachy2 = ["reachy2_sdk>=1.0.15,<1.1.0"]
# Seeed Studio reBot B601-DM follower (motorbridge / CAN) + StarArm102 / reBot Arm 102
# leader (motorbridge-smart-servo / FashionStar UART servos).
rebot = ["lerobot[motorbridge-dep]", "lerobot[motorbridge-smart-servo-dep]"]
kinematics = ["lerobot[placo-dep]"]
intelrealsense = [
"pyrealsense2>=2.55.1.6486,<2.57.0 ; sys_platform != 'darwin'",
@@ -195,7 +213,7 @@ groot = [
sarm = ["lerobot[transformers-dep]", "pydantic>=2.0.0,<3.0.0", "faker>=33.0.0,<35.0.0", "lerobot[matplotlib-dep]", "lerobot[qwen-vl-utils-dep]"]
xvla = ["lerobot[transformers-dep]"]
eo1 = ["lerobot[transformers-dep]", "lerobot[qwen-vl-utils-dep]"]
hilserl = ["lerobot[transformers-dep]", "gym-hil>=0.1.13,<0.2.0", "lerobot[grpcio-dep]", "lerobot[placo-dep]"]
hilserl = ["lerobot[transformers-dep]", "lerobot[dataset]", "gym-hil>=0.1.13,<0.2.0", "lerobot[grpcio-dep]", "lerobot[placo-dep]"]
# Features
async = ["lerobot[grpcio-dep]", "lerobot[matplotlib-dep]"]
@@ -261,6 +279,7 @@ all = [
"lerobot[lekiwi]",
"lerobot[openarms]",
"lerobot[reachy2]",
"lerobot[rebot]",
"lerobot[kinematics]",
"lerobot[intelrealsense]",
"lerobot[diffusion]",
@@ -306,6 +325,20 @@ lerobot-annotate="lerobot.scripts.lerobot_annotate:main"
lerobot-rollout="lerobot.scripts.lerobot_rollout:main"
# ---------------- Tool Configurations ----------------
# cu128 wheels keep broad hardware reach; the driver floor is 570.86.
# To use a different CUDA variant, reinstall torch with an explicit index, e.g.:
# uv pip install --force-reinstall torch torchvision \
# --index-url https://download.pytorch.org/whl/cu130
[[tool.uv.index]]
name = "pytorch-cu128"
url = "https://download.pytorch.org/whl/cu128"
explicit = true
[tool.uv.sources]
torch = [{ index = "pytorch-cu128", marker = "sys_platform == 'linux'" }]
torchvision = [{ index = "pytorch-cu128", marker = "sys_platform == 'linux'" }]
[tool.setuptools.package-data]
lerobot = ["envs/*.json", "annotations/steerable_pipeline/prompts/*.txt"]