The benchmarks/streaming harness (matrix submitter, summarizer, decode
diagnostic) and the robocasa SLURM scripts are cluster-specific tooling,
not part of the streaming feature. The example's --dummy mode covers
throughput measurement for reviewers. Recoverable from git history
(894fc6bfb) for cluster runs. Example docstring de-personalized.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The custom episode pool becomes a pure `datasets` pipeline:
split_dataset_by_node -> batch(by_column="episode_index")
-> shuffle(buffer=episode_pool_size) # episode pool
-> map(explode + exact delta windows) # episode -> frames
-> shuffle(buffer=frame_shuffle_buffer_size) # frame interleave
and the torch IterableDataset wrapper keeps only per-sample video decode
(decode-on-exit), image transforms, task lookup, and decode/fetch timing.
Replaced by native machinery and deleted: the pooled-episode admission
loop, the refcounted video prefetcher, manual worker shard striding plus
the worker-split suppression patch, the per-(epoch, rank) shard-order
permutation, the per-consumer SplitMix64 RNG, and fast-forward resume.
DataLoader workers are split by `datasets` itself; .shuffle() permutes
shard order per epoch natively; resume delegates to the native
state_dict/load_state_dict (exact with num_workers=0; with workers use
torchdata's StatefulDataLoader, which checkpoints per-worker state
through the same protocol). An in-flight epoch counter ensures a
mid-iteration state_dict records the epoch the stream position belongs
to. Buffer contents are skipped on resume (documented datasets
behavior): never repeats data, drops at most ~pool + frame-buffer frames.
Randomness is unchanged: a batch still mixes up to episode_pool_size
episodes; delta windows are still exact in-episode slices with correct
boundary padding (value-verified against the map-style dataset). The
known trade accepted with this rewrite: no video prefetch-on-admit, so
remote decode pays per-frame range reads at yield time - use a colocated
bucket (data_files_root) at large scale.
The delta-consistency tests gained a scalar-comparison branch: they
silently skipped python-scalar keys before (stale `check` variable),
exposed by the new pipeline's key ordering.
Requires datasets with #8259 (pinned to the merge commit on this
branch). Example updated to per-rank native resume via torchdata's
StatefulDataLoader when available.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The native streaming pipeline calls .shuffle() on top of batch(by_column=...),
which crashes on released datasets 5.0.0 (batch-accumulator flag dropped on
shard/shuffle re-creation). The fix (datasets#8259) is merged but unreleased,
so pin datasets to the merge commit 2c45eab on this branch via [tool.uv.sources].
Drop this pin and bump the floor in `dependencies` once the next datasets
release ships the fix.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Shards were assigned to consumers in file-index order, so a sub-epoch
run over a corpus consolidated source-by-source trains on whatever the
first N% of files contains and drifts curriculum-style as sources change
under it. Permute the rank's shard list with a seeded RNG before worker
striding: a 30%-of-epoch run now sees a uniform 30% sample of files.
The permutation is seeded by (seed, epoch, rank) only - every DataLoader
worker of a rank must derive the identical list, since workers stride it
and disagreement would create overlapping shard assignments. It re-draws
each epoch, is the identity when shuffle=False, and stays deterministic
for fast-forward resume.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Allow datasets 5.x (pin >=4.7,<6; lockfile moves to 5.0.0) and use its
Arrow-native batch(by_column="episode_index") (huggingface/datasets#8194
sibling, #8172) for episode admission when available - one Arrow
accumulation per episode instead of one Python dict per row - with the
existing row loop as the 4.x fallback. A parity test asserts both paths
group identically.
Also fixes a latent worker bug this surfaced: `datasets` detects torch
DataLoader workers and re-splits its shards internally (_iter_pytorch),
on top of our explicit per-worker shard assignment. That second split
silently drops data whenever a per-worker stream has fewer internal
shards than there are workers (masked so far by single-file test
fixtures), and on datasets 5.0 it crashes by_column batching outright.
The worker context is now hidden from `datasets` while draining streams
we already partitioned (process-local patch, restored on exit).
The multi-shard shuffle buffer (huggingface/datasets#8194) is
intentionally NOT used: frame-level shuffling upstream of episode
grouping would fragment episodes and break delta windows. Its threaded
multi-source prefetch idea remains a follow-up for episode admission if
fetch timings warrant it.
Verified on both datasets 4.8.5 (fallback) and 5.0.0 (native): 27/27
streaming tests each; full datasets suite 469 passed under 5.0.0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The remote commit (2ab71231c) added an opt-in episode pool, deferred
decode in the legacy buffer path, decode/fetch timing instrumentation,
remote-IO retries (video_utils), and 32MB row-group writing
(dataset_tools). The pool rewrite on this side makes the episode pool
the only iteration path (with prefetch-on-admit, per-consumer seeding,
worker-exact fast-forward resume), so streaming_dataset.py resolves to
the rewrite with the remote instrumentation ported into it:
- 5-slot shared counters + timing_stats() (decode_s_total/fetch_s_total)
- fetch timed around episode admission, decode timed around emission
- benchmark/slurm keep the remote updates, with episode_pool_size as the
knob (buffer_size deprecated and ignored)
video_utils retries and dataset_tools row groups are taken unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The fast-forward skip assumed every DataLoader worker delivers batches;
workers that own no shards yield nothing and are stopped, so the batch
round-robin runs over min(num_workers, num_shards) active workers. Use
that effective count (shard-less workers skip nothing). Adds a resume
test under num_workers=2 asserting exact continuation.
Note: the test fixtures write a single parquet file regardless of
data_files_size_in_mb, so worker-splitting tests exercise the degenerate
single-shard layout; multi-shard behavior is covered by the rank-level
split_dataset_by_node tests.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Replace the shard/Backtrackable/decoded-shuffle-buffer internals with an
episode pool: each (rank x worker) consumer keeps episode_pool_size whole
episodes' tabular rows in RAM and emits uniformly random frames across
them. delta_timestamps windows become exact in-RAM slices with correct
boundary padding (the Backtrackable machinery and its lookback/lookahead
ceilings are gone), and video is decoded only when a sample is emitted,
so pool memory stays tabular-sized instead of buffer_size decoded
samples.
- Prefetch-on-admit: when streaming from a remote source, each pooled
episode's video files download to a local cache in the background
(refcounted, since v3 packs several episodes per file; deleted on
eviction), so decode-on-exit reads local bytes instead of paying
network seek latency.
- Per-consumer RNG derived from (seed, epoch, rank, worker): consumers
decorrelated, runs reproducible, epochs reshuffle automatically.
- Deterministic fast-forward resume: load_state_dict takes the trainer's
{batches_consumed, batch_size}; each worker re-derives its own skip
from the DataLoader's round-robin batch assignment and replays
tabular-only (no decode). Exact within an epoch, works with
num_workers > 0, and the same state file serves every rank. Replaces
the per-shard HF state_dict approach, which lived in worker processes
and could not be captured from the trainer.
- Shard-cap default removed (max_num_shards=None uses every parquet
shard); runtime warnings for non-divisible world sizes (datasets
degrades to read-everything splitting) and workers left without
shards.
- episode_pool_size replaces buffer_size (deprecated, ignored with a
warning); decoder cache sized to the pool working set, capped at 128.
Legacy order-replication tests asserted the old buffer algorithm
step-by-step and are rewritten as behavior contracts (exactly-once
coverage, per-seed determinism, epoch reshuffle). Value-level parity
tests against the map-style dataset pass unchanged.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The dataset is already rank-disjoint via split_dataset_by_node;
accelerate's IterableDatasetShard wrapper kept only every Nth batch of
each rank's stream, silently training on 1/N of the data per pass while
decoding all of it. The --dummy benchmark path never prepared the
loader, so benchmarks were unaffected.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- streaming_dataset: defer torchcodec decode until a sample leaves the shuffle
buffer (buffer now holds ~KB tabular rows, not MB of pixels) and add an opt-in
episode-pool shuffle (episode_pool_size) with exact in-episode delta lookups;
expose decode/fetch timing_stats.
- video_utils: retry transient hf:///fsspec/httpx transport errors during
streaming decode (LEROBOT_REMOTE_IO_MAX_RETRIES).
- dataset_tools: write multiple ~32MB row groups with a page index to bound
per-shard streaming memory.
- benchmarks/slurm: streaming benchmark + matrix submitter updates.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(datasets): expose a generator on EpisodeAwareSampler for distributed shuffle sync
In distributed training, accelerate can only synchronize the shuffle
permutation across ranks when the sampler exposes a generator attribute.
EpisodeAwareSampler shuffled via the global torch RNG, so disjoint batch
shards relied on every rank's global CPU RNG staying in lockstep forever;
any rank-asymmetric RNG consumption (e.g. eval rollouts on the main
process only) silently desynced the permutations and ranks trained on
overlapping/missing samples.
* fix(train): seed sampler generator and gate dataset download per node
- Pass a generator seeded with cfg.seed to EpisodeAwareSampler so
accelerator.prepare registers it as the synchronized RNG and the
shuffle order is reproducible.
- Gate the initial make_dataset call on is_local_main_process instead of
is_main_process: the global main process only exists on node 0, so on
every other node all local ranks were downloading the dataset and
building the Arrow cache concurrently.
The data_files_root/bucket path reads an unversioned source (e.g. `main`), which can
carry extra annotation columns not in the dataset's feature contract — notably
`language_events`, a variable-length list (length 0..N per frame). Passed through to the
sample, these break default DataLoader collation ("each element in list of batch should
be of equal size"), which is why bucket jobs failed while the hub path (pinned to the
clean v3.0 revision) succeeded.
Drop any hf_dataset column not in meta.features after load. No-op on a clean revision;
removes language_events/language_persistent on main. Verified by reproducing the bucket
code path locally via --data_files_root hf://datasets/<repo> (parquet builder + main
columns): now decodes and collates instead of raising.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
So the A100/H100 no-AV1-NVDEC limitation applies to most LeRobot v3 datasets, not just
RoboCasa — GPU decode needs an Ada GPU, an hevc/h264-encoded dataset, or a re-encode.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
NVIDIA's decode support matrix: the compute GPUs A100 (GA100) and H100 (GH100) have no
AV1 NVDEC decoder; only Ada (L4/L40/RTX40) and some Ampere (A10/A40/A16) do. So on
A100/H100 nodes, AV1 datasets must be decoded on CPU or re-encoded to H.265/H.264 — no
torchcodec build enables cuda AV1 decode there. Also distinguish that error from
"Unsupported device: cuda (variant: ffmpeg)", which is a torchcodec-built-without-CUDA
issue. Update diagnose_decode.py message + benchmark README accordingly.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add CONDA_ENV=<name> to run each matrix job via `conda run --no-capture-output -n
<name>` — works inside the dash `sbatch --wrap` without sourcing conda.sh / activating,
and streams logs live. Point it at a conda env with a modern torchcodec (>=0.11) +
datasets (>=4.7); the default cluster `base` env is often too old to decode AV1.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- benchmark: raise SystemExit if 0 frames were measured, so a run that produces no
batches (swallowed decode error, all batches dropped) fails loudly instead of being
reported green with NaN/zero numbers (the misleading "COMPLETED" CUDA jobs).
- add benchmarks/streaming/diagnose_decode.py: isolates the streaming decode path
(resolve path -> fsspec.open -> torchcodec VideoDecoder -> get one frame) and prints
package versions + the first bytes of the handle. Pinpoints decode failures: bad/
placeholder bytes vs ffmpeg/torchcodec build issue. RoboCasa videos are AV1; the
failure message calls out AV1 decoder + NVDEC-on-Ada requirements explicitly.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
`sbatch --wrap` runs the wrapped body under /bin/sh (dash), which has no
`set -o pipefail`, so every matrix job died on line 1 ("Illegal option -o pipefail")
before reaching the benchmark. The command has no pipes, so drop the bashism and chain
with `&&` (cd-guards the run) — fully POSIX-sh compatible. Runtime env expansion
(${HF_HOME:-$SCRATCH/hf_home}) is preserved.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
For a bandwidth-sensitive benchmark, concurrent jobs would share the network to the
Hub/bucket and corrupt throughput numbers. Chain the matrix jobs with
--dependency=afterany (captured via `sbatch --parsable`) so SLURM runs exactly one at a
time while keeping each config an isolated job (own log + per-job OOM reporting).
afterany keeps the chain going if one job fails/OOMs. SERIAL=0 restores parallel
submission for OOM-isolation-only testing.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
slurm/run_streaming_matrix.sh fans the benchmark matrix (sources {hub,bucket,
warmed_bucket} x modes {single,sarm} x decode {cpu,cuda}) out as isolated single-GPU
SLURM jobs, so an OOM in one config is contained and reported per-job by SLURM. Worker
count and shuffle buffer are bounded (lower for cuda, which holds a CUDA context + NVDEC
session per worker) to avoid host/VRAM OOM. Source/mode/decode/workers/buffer/account/
partition are env-overridable; SOURCES/MODES/DECODES select subsets.
benchmarks/streaming/summarize_results.py collapses the per-run JSONs into one comparison
table + summary.csv (frames/s/node, first-batch + p50/p95/p99 latency, cache hit-rate).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add `video_decode_device` to StreamingLeRobotDataset and a `device` arg to
VideoDecoderCache, passed to torchcodec's VideoDecoder. "cuda" offloads H.264/H.265
decode to the GPU's dedicated NVDEC engine (independent of the training SMs); requires
a CUDA-enabled torchcodec build.
benchmark: `--video_decode_device` flag. With cuda + num_workers>0 it forces the
`spawn` start method (CUDA cannot init in forked workers) and disables CPU pin_memory
(frames are already on-GPU). Decode device is recorded in results and the output
filename. README documents the NVDEC option and its concurrency/IPC caveats.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- benchmark: frames_per_s_node now measures sustained wall-clock throughput over the
post-warmup window. The previous metric summed inter-batch gaps, which collapse to ~0
under async prefetch (consumer drains a pre-filled queue) and overstated throughput ~100x.
- VideoDecoderCache gains an optional shared [hits, misses, evictions] counter tensor;
StreamingLeRobotDataset.video_decoder_cache_stats() aggregates it across DataLoader
workers (lock-free, approximate; hit_rate preserved). Fixes empty cache stats with workers.
- StreamingLeRobotDataset.data_files_root: read bulk data/ + videos/ from an fsspec root
(e.g. hf://buckets/<owner>/<name>) while metadata still loads from repo_id. Enables
bucket / prewarmed-bucket benchmark sources without copying metadata. Exposed as
benchmark --data_files_root.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
make_frame used `item["index"] / fps` (a dataset-global value) as the in-file
video timestamp. That only matches the file timeline when the whole dataset is a
single video (as in the test fixtures); on multi-file v3 datasets it decodes
out-of-range frames and crashes (e.g. RoboCasa: "Invalid frame index=23314614 ...
must be less than 41021").
Mirror the map-style reader: use the episode-local `timestamp` column as the base,
clamp delta query timestamps to per-camera episode-local bounds [0, duration], and
shift by the episode's `from_timestamp` per camera at decode time. For single-file
datasets `from_timestamp + timestamp == index / fps`, so existing parity tests are
unaffected; multi-file streaming is now correct.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- examples/scaling/train_streaming_multinode.py: Accelerate-based distributed/
resumable streaming training (no DistributedSampler; rank/world_size auto-resolved),
checkpoints the dataset stream state, and supports a --dummy pure-dataloading path
with throughput logging. SLURM launcher in slurm/train_streaming_robocasa.sh.
- benchmarks/streaming/benchmark_streaming.py: dummy-consumer dataloading benchmark
(single / sarm frame modes) emitting frames/s/node, p50/p95/p99 sample latency,
first-batch latency, and VideoDecoderCache reuse stats as JSON + CSV. SLURM launcher
+ README documenting the source/node/mode matrix and manual bucket prewarming.
- VideoDecoderCache: add hit/miss/eviction counters and a stats() method so the
benchmark can surface decoder thrash (no new cache, no eviction-policy change).
- tests/datasets/test_streaming_distributed.py: accelerate-launch smoke test asserting
per-rank disjointness; skips (does not false-pass) when <2 processes spawn.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add the large-scale streaming pieces that were missing from the frame-streaming
internals, keeping the existing Backtrackable + output-reservoir frame-shuffle:
- split_dataset_by_node(rank, world_size) before the per-shard loop so each rank
streams a disjoint set of shards (fixes duplicate data across GPUs). rank and
world_size auto-resolve from Accelerate state / RANK,WORLD_SIZE env / (0, 1).
- get_worker_info() shard splitting so DataLoader workers within a rank don't
yield duplicate frames.
- Dynamic Backtrackable window (dynamic_bounds=True) sized to the requested
delta_timestamps, removing the fixed 100-frame ceiling so long horizons (e.g. a
SARM window ~160 frames) reach real frames instead of silently padding. Fix the
peek_back off-by-one: history = lookback + 1.
- video_decoder_cache_size knob; default (active_shards + 1) x num_cameras so the
live decoder working set does not thrash the VideoDecoderCache LRU.
- state_dict()/load_state_dict() for resume (per-shard HF stream state + exhausted
set + RNG). Reservoir is re-warmed, so resumption is not bit-exact (documented).
- factory.py wires buffer_size from a new DatasetConfig.streaming_buffer_size field
instead of repurposing max_num_shards as the worker count.
Tests: tests/datasets/test_streaming_native.py covers distributed disjointness,
worker de-duplication, the SARM-length window, resume, schema parity vs map-style,
local video path resolution, and shuffle decorrelation. 21 passed (13 existing + 8).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
* feat(processor): add in-memory pipeline serialization
Expose processor pipeline config and tensor state without requiring temporary files, so processors can be transported, compared, or hashed directly in memory.
* feat(processor): enhance DataProcessorPipeline with registry support
- Added a new RegisteredLazyTensorStateStep for registry-based serialization tests.
- Improved state filename handling in _get_state_filename method.
- Refactored validation logic in _validate_loaded_config to simplify parameter types.
- Updated tests to verify registry step functionality and ensure correct state loading.
* refactor(processor): update state handling in DataProcessorPipeline
- Introduced a new static method _get_state_key to derive in-memory state keys from serialized filenames.
- Updated state_dict and load_state_dict methods to use suffixless state keys instead of filenames.
- Adjusted related tests to reflect changes in state key handling, ensuring consistency in state management
* fix(processor): update loaded_config argument description in DataProcessorPipeline
- Clarified the documentation for the loaded_config parameter to indicate that it may be a non-dictionary value, enhancing understanding for future developers.
* first commit
* feat(policies): add VLA-JEPA
* feat(policies): add VLA-JEPA
* support vla_jepa
* (feat)policies: add VLA-JEPA
* linting
* adding deps to pyproject.toml
* updating uv lock
* adding guards to avoid needing transformers and diffusers for type checking and basic tests
* fixing action and state dim
* fix warnings with qwen processor kwargs
* fixing wm_loss not propagating
* adjusting obs steps, tublets size to match original implementation
* some more fixes to be closer to the original implem
* adding more tests to ensure good coverage
* align VLA-JEPA architecture with original checkpoint
- Remove stale `action_num_heads` / `action_attention_head_dim` config fields;
DiT head dimensions are now always derived from the preset (DiT-B/L/test).
- Add `num_target_vision_tokens` and `action_max_seq_len` config fields required
by the action head's future-token embedding and positional embedding tables.
- Fix default `qwen_model_name` to 2B (matches all released checkpoints).
- Rename `ActionEncoder` attrs w1/w2/w3 → layer1/layer2/layer3 to match
checkpoint key names; replace `nn.Sequential` decoder/state-encoder with
`_MLP2` (layer1/layer2 naming).
- Fix `VLAJEPAActionHead` to size ActionEncoder and StateEncoder at `inner_dim`
(DiT input width) rather than `action_hidden_size` (DiT output width).
- Rename `DiT.blocks` → `transformer_blocks` and `attn` → `attn1` to match
checkpoint; add alternating cross/self attention (even blocks cross-attend to
Qwen context, odd blocks self-attend).
- Add `DiT-test` preset for unit tests.
- Rewrite `ActionConditionedVideoPredictor` with explicit ViT-style blocks
(`_PredictorBlock` with fused qkv) to match checkpoint structure; rename
`encoder`/`norm`/`proj` → `predictor_blocks`/`predictor_norm`/`predictor_proj`.
* propagate action_is_pad masking through VLA-JEPA policy pipeline
Pass the `action_is_pad` tensor from the batch through to the action head
so padded timesteps are excluded from the flow-matching loss.
* update VLA-JEPA tests for arch changes and action_is_pad
- Switch conftest to use `action_model_type="DiT-test"` now that
`action_num_heads` / `action_attention_head_dim` have been removed.
- Add action_head tests covering fully-padded loss (zero) and equivalence
of action_is_pad=None vs all-zeros mask.
- Remove obsolete `test_native_to_lerobot_wm_only` test.
* add VLA-JEPA documentation
Covers architecture overview, pretrained checkpoints, config reference,
training/eval commands for LIBERO-10, and guidance on fine-tuning for
single-camera datasets.
* add one-shot script to convert ginwind/VLA-JEPA checkpoints to safetensors (will remove once migrated)
* make default params more aligned with paper and pretrained models
- adding possibility of freezing qwen backbone and world model
- added tests for weight loading
* trying out to re-init the action head to avoid pretraining dimension mismatch
* allow different state dim and action dim
* removing missleading future_action_window_size to just use chunk_size
* lots of changes to make existing weights work, need to massively refactor the pre and post processing
* refactoring into using pre and post processor
* pre-commit cleanup
* fixing doc defaults args
Signed-off-by: Maxime Ellerbach <maxime@ellerbach.net>
* adressing dtype zeros issue
* adding guard for diffusers
* fixing training and exal examples
* trying to close success rate gap
* fix qwen norm layer output libero eval is now as expected
* adding instructions for different embodiement + fixing some tests
* smol fix to avoid having default CPU device when training
* fixing misconception about multiview / singleview handling
* removing conversion script
* adding licences
* adding .mdx docs and shortening polivy_vla_jepa_README.md
* removing useless pre-processor
* cleanup
* removing swish in favor of silu
* adding configuration gripper index and threshold
* fixing simlink
---------
Signed-off-by: Maxime Ellerbach <maxime@ellerbach.net>
Co-authored-by: ginwind <ginwind@mail.ustc.edu.cn>
The docs pointed at src/lerobot/datasets/v30/, which does not exist.
Both scripts actually live in src/lerobot/scripts/:
- convert_dataset_v21_to_v30.py
- augment_dataset_quantile_stats.py
Updated the four references (one python -m module path and three
file-path invocations) to the correct location, matching each
script's own usage docstring.
* fix(train): enable relative action overrides for pretrained processors
Keep pretrained processor pipelines when use_relative_actions is enabled and
apply relative/absolute action processor settings through overrides. Rename the
relative action processor registry key to relative_actions_processor.
* fix(config): reject rename_map without pretrained checkpoint
Fail fast when rename_map is set during fresh initialization, since fresh
configs derive feature names from the current dataset and no rename is applied.
---------
Co-authored-by: Pepijn <138571049+pkooij@users.noreply.github.com>
* feat(rewards): add TOPReward reward model
* refactor(rewards): clean up TOPReward processor/model
* fix(rewards/topreward): add missing input keys mm_token_type_ids
* fix(rewards/topreward): fix pyproject extra typo and simplify processor (#3653)
Add lerobot[topreward] extra to all in
pyproject.toml, drop the redundant labels arg in scoring, and
collapse the dead-branch shape check in the encoder processor.
* optmize topreward input processing (#3660)
---------
Co-authored-by: Cole <91766445+jcoleharrison@users.noreply.github.com>
Co-authored-by: Haoming Song <haomingsong24@gmail.com>
PR #3145 added YAML support for policy.path but left two bugs:
1. extract_path_fields_from_config only deleted config_data[field] when
no sibling overrides existed. With siblings, the dict stayed in place
and draccus crashed decoding it as PreTrainedConfig (no 'type' key).
Sibling overrides go into _config_yaml_overrides and are applied later
by from_pretrained(), so the field can always be removed.
2. wrap() updated config_path_cli to the cleaned temp file path but
never propagated it to the draccus.parse fallback branch. cli_args
still contained --config_path=<original>, so draccus read the
original YAML with path: still present.
Tests passed because they (a) called extract_path_fields_from_config
directly and (b) included type: alongside path: in the YAML, sidestepping
both bugs.
Co-authored-by: Steven Palma <imstevenpmwork@ieee.org>
* fix(deps): cap placo below 0.9.16 and harden kinematics import
placo 0.9.16 links against liburdfdom_sensor.so.4, which is unavailable
on Ubuntu 24.04 (noble ships urdfdom 3.x). Importing placo on that base
crashes with:
ImportError: liburdfdom_sensor.so.4.0: cannot open shared object file
This broke nightly Latest Deps tests (CPU and GPU) when the lockfile
upgrade picked placo 0.9.16, since lerobot.model.kinematics
unconditionally imports placo when _placo_available is true, and that
check (importlib.util.find_spec) cannot detect dlopen failures of
transitive shared libraries — so unrelated subsystems (RL actor,
gym_manipulator) became unimportable.
Two changes:
1. Pin placo to <0.9.16 in pyproject.toml + regenerate uv.lock
(0.9.16 → 0.9.15). Short-term unblock for nightly CI until system
urdfdom 4.x is broadly available.
2. Harden the import guard in src/lerobot/model/kinematics.py:
wrap 'import placo' in try/except ImportError so a missing
transitive .so no longer crashes module import. RobotKinematics
instantiation now raises an informative ImportError citing the
underlying dlopen failure via _raise_if_placo_unusable().
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(kinematics): hoist _placo_runtime_error to module scope for mypy
Mypy walks the TYPE_CHECKING branch in which the runtime else-block is
not executed, so _placo_runtime_error was only defined at runtime and
mypy reported 'Name "_placo_runtime_error" is not defined' on the
three references inside _raise_if_placo_unusable. Declare the symbol
unconditionally at module scope with a default of None; the runtime
import-failure branch still assigns to it.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* style(kinematics): drop verbose comments around placo import guard
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* chore(gr00t): sync with #3606 for fixing gr00t config crash
* fix(pi0&pi05): fix graph break caused by deepcopy of past_key_values in sample_actions
* fix(pi0&pi05): fix frequent recompile caused by compute_layer_complete
* feat(test): add compile test and benchamrk for pi0 and pi05
* feat(test): add comprehensive testing for pi0 and pi05. Including processor, forward, sample action, etc.
- Fixed broken API examples in Lerobot Imitation Learning Documentation
- Teleoperation with cameras improved by adding a fixed frequency in the loop (without it the cameras feed gets very slow)
- Wrapped record example script in main() to avoid problems on Mac
- Previously teleoperation example was using SO-ARM and teleoperation with cameras was using Koch. I changed it to use SO-ARM in all of the examples.
- Added section on how to train with HF Jobs - CLI and Python examples
- Replaced lerobot-record with lerobot-rollout in policies examples
VideoDecoderCache used an unbounded dict keyed on absolute path, with no
eviction in the standard LeRobotDataset path. With shuffled iteration over
datasets that have many distinct mp4 files, every DataLoader worker
accumulated one cached (VideoDecoder, fsspec file handle) pair per distinct
path it had ever touched. Per-entry cost is ~3-5 MB of host RAM plus one
open FD; at ~8 k entries this is roughly 30 GB per worker.
This was hit in the wild during a SmolVLA training run on a 4,195-episode
SO-101 dataset (8,390 mp4s, two cameras per episode). dmesg showed
anon-rss climbing to 34.9 GB on a single pt_data_worker before the OOM
killer fired ~30 min into training; with --num_workers=8 the per-worker
peak halved to 17.9 GB, which is the expected inverse-scaling signature
when the leak is per-decode and the workload is split across workers. The
working workaround on the affected platform was --dataset.video_backend=pyav,
because the pyav path opens/closes per call and never touches this cache.
Switch the backing store to an OrderedDict and evict LRU entries when the
cap is reached, closing the evicted file handle inside the lock so we do
not leak FDs either. Default cap is DEFAULT_DECODER_CACHE_SIZE = 100,
overridable via LEROBOT_VIDEO_DECODER_CACHE_SIZE or by passing max_size=
to the constructor; max_size=None restores the legacy unbounded behaviour
for callers that need it.
Validation on the original failing workload (decode_video_frames_torchcodec
called over real mp4s from the affected SO-101 dataset):
unbounded: 300 files -> +1087 MB host RSS, cache=300, still climbing
cap=50: 500 files -> +266 MB host RSS, cache=50, stable
cap=50: 2000 calls -> +312 MB host RSS, cache=50, stable
cap=100: 1000 calls -> +470 MB host RSS, cache=100, stable
Three independent seeded runs at cap=50 agreed to within 1% (263 / 266 /
265 MB delta), and the 2000-call multi-pass run shows RSS plateaus after
the cap is reached instead of drifting.
Tests in tests/datasets/test_video_decoder_cache.py cover:
default-is-bounded, size cap, LRU ordering, FD close on eviction, FD close
on clear(), cache-hit invariance, max_size=None fallback, and env-var
override. No regressions in test_video_encoding.py, test_streaming.py, or
test_dataset_reader.py (73 prior tests still pass alongside the 8 new ones).
* feat(utility): adding video re-encode utility
* feat(edit): adding a new lerobot-edit-dataset tool to re-encode all the videos of a dataset
* chore(format): formatting code
* chore(review): fix Claude reviews
* test(reencode dataset): adding missing test for reencode dataset
* Add extensive language support
* Address review: split persistent/event schemas, drop event timestamps
- recipe.py: derive _VALID_ROLES/_VALID_STREAMS from MessageRole/MessageStream Literals
- dataset_metadata.py: keep CODEBASE_VERSION at v3.0
- language.py: remove RESERVED_STYLES; split arrow/feature schemas into
persistent (with timestamp) and event (without timestamp); add docstrings
- language_render.py: events use frame-row timestamp implicitly; no
per-event timestamp filtering or sorting
- converters.py: drop unused subtask_key passthrough
- add docstrings to new public APIs (recipe, render_messages_processor, collate)
- update tests for split schemas; revert uv.lock
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Add docstrings to all new helpers; revert uv.lock
Covers private helpers in recipe.py, language.py, language_render.py,
and render_messages_processor.py. Also reverts uv.lock to main (it was
re-generated by `uv run` during local checks).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(language): add motion (persistent) and trace (event-only) styles
Promote the previously-reserved motion/trace styles to first-class core
styles. motion routes to language_persistent (it tracks robot state over
time); trace routes to language_events (single-moment annotations).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(language): per-camera tagging on view-dependent styles
Adds a nullable `camera` field to the language row struct (both persistent
and event variants) so view-dependent styles like `vqa` can carry which
`observation.images.*` view they were grounded against. Without this,
multi-camera datasets ended up with multiple `(vqa, role)` rows at the
same timestamp that the resolver could not disambiguate.
- `language.py`: add `camera` to PERSISTENT_ROW_FIELDS / EVENT_ROW_FIELDS,
to both Arrow struct types and the HF datasets feature mappings;
introduce VIEW_DEPENDENT_STYLES = {vqa, motion, trace} plus
`is_view_dependent_style` and `validate_camera_field` helpers (camera
required iff style is view-dependent).
- `language_render.py`: thread an optional `camera=` kwarg through every
resolver (`active_at`, `emitted_at`, `nth_prev`, `nth_next`) and through
`_matching_rows` / `_select_*`, so recipes can disambiguate per-camera
VQA with `emitted_at(t, style=vqa, role=assistant, camera=...)`.
Without a `camera` filter, multi-row matches keep raising the existing
ambiguity error — which is the desired behaviour on multi-camera data.
- `recipes/pi05_hirobot.yaml`: replace the single `ask_vqa` branch with
`ask_vqa_top` and `ask_vqa_wrist` per-camera sub-recipes (each carrying
the matching image block), keeping the original 0.20 budget and
documenting the customization point for datasets with different cameras.
- Tests: schema test asserts the new field order; new tests cover
`is_view_dependent_style`, `validate_camera_field` (both required and
forbidden directions), per-camera `emitted_at` filtering, and the
ambiguity error when two cameras emit `(vqa, assistant)` at the same
timestamp without a `camera=` filter. RenderMessagesStep + dataset
passthrough fixtures updated to include the new field.
- `docs/source/language_and_recipes.mdx`: document the `camera` field,
the per-camera resolver pattern, and the canonical recipe convention.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(language): drop motion from VIEW_DEPENDENT_STYLES
Motion primitives are described in robot-frame (joint / Cartesian) terms,
not pixel space, so they are camera-agnostic. Only `vqa` (event) and
`trace` (event, pixel-trajectory) are view-dependent.
The `camera` field stays on PERSISTENT_ROW_FIELDS for schema symmetry —
the validator, resolver, and HF feature mapping behave identically across
the two columns regardless of which styles populate `camera` today —
but persistent rows now always have `camera=None` in practice.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(language): task_aug style + automatic ${task} rephrasing rotation
Adds task-prompt diversity (Xiao 2022 / CAST) without touching
``meta/tasks.parquet`` or forcing recipes to opt in. The plan reserved
``task_aug`` as a future style; this lands it now.
- ``language.py``: add ``task_aug`` to ``CORE_STYLES`` and
``PERSISTENT_STYLES``. ``column_for_style("task_aug")`` returns
``language_persistent`` so PR 2 writers route it correctly.
- ``language_render.py``: ``_resolve_task`` now consults the persistent
slice for rows of ``style="task_aug", role="user"``. When any exist
it picks one deterministically by ``sample_idx`` (blake2b-keyed, not
Python's randomized hash) so an epoch sees every rephrasing of every
episode while the same sample still resolves identically across
reruns. Falls back to the canonical ``meta/tasks.parquet`` task when
no rephrasings are present, so existing datasets and unannotated runs
keep their behaviour. Explicit ``task=`` overrides still win.
- Tests: rephrasing coverage across samples, determinism on repeat
``sample_idx``, fallback when persistent has no ``task_aug`` rows,
and explicit override priority.
Recipes get this for free: any ``${task}`` placeholder rotates through
the available rephrasings. Recipes that want the literal canonical task
can override the binding.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(language): tool catalog in meta/info.json + LeRobotDatasetMetadata.tools
Stores OpenAI-style function schemas at ``meta/info.json["tools"]`` so
datasets can declare which tools are available (today: just ``say``;
tomorrow: per-dataset extensions). The ``DEFAULT_TOOLS`` constant
fills in for unannotated datasets so chat-template consumers don't
have to special-case anything.
Three pieces:
- ``language.py``: ``SAY_TOOL_SCHEMA`` and ``DEFAULT_TOOLS``
constants. Single source of truth — PR 2's writer and PR 3's
runtime tool registry will both import from here instead of
duplicating the dict.
- ``dataset_metadata.py``: ``LeRobotDatasetMetadata.tools`` property
reads ``info.json["tools"]`` and falls back to ``DEFAULT_TOOLS``.
Returns deep-copied dicts so callers can mutate the result safely.
- ``docs/source/tools.mdx``: spec page covering the catalog, per-row
invocations, and the three-step "how to add a new tool" workflow
(declare schema, implement, register). Linked from the docs
toctree under the Datasets section.
This lays the groundwork for PR 2's pipeline writing the catalog out
during annotation, and PR 3's ``src/lerobot/tools/`` package shipping
runnable implementations (one file per tool — first up:
``say.py`` wrapping Kyutai's pocket-tts).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* Apply ruff and prettier formatting after merge
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* refactor(language): unify resolver dispatch and prune redundant test scaffolding
* Drop the unused `events` kwarg from `active_at`/`nth_prev`/`nth_next`;
only `emitted_at` actually consults events. The dispatcher in
`_resolve_spec` now passes events conditionally.
* Replace the dual `_persistent_sort_key`/`_event_sort_key` pair with a
single `_row_sort_key` and drop the `sort_key` parameter from
`_select_one`. Event rows lack `timestamp` (it is implicit in the
frame) and now default to `0.0` for sort purposes — the
`(style, role)` tiebreaker is unchanged.
* Inline `_select_latest` into `active_at` (its only caller).
* Collapse `emitted_at`'s dual-branch into one `_select_one` call.
* Tighten `_validate_persistent_resolver` to a single
`column_for_style(style) != LANGUAGE_PERSISTENT` check.
* Parameterize `test_per_camera_blend_renders_both_views` over the two
cameras and factor the sub-recipe builder into `_vqa_subrecipe` so
the test no longer hand-rolls two near-identical recipe blocks.
Net -98 LOC; behavior, public resolver names, and test expectations
unchanged.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(language): always raise on ambiguous resolver matches
`_select_one` previously skipped its ambiguity check whenever any of
`role`/`tool_name`/`camera` was set, on the assumption that the caller
had already pinned down a unique row. That left a real ambiguity hole
for VQA: with two cameras emitting `(vqa, assistant)` at the same
frame, `emitted_at(..., role="assistant")` silently picked the first
sorted row instead of telling the recipe to add `camera=...`. The
existing `test_emitted_at_raises_on_ambiguous_per_camera_vqa` test
already encoded the desired behavior.
Tighten the check: any time `len(rows) > 1` we now raise with the
selectors echoed back, so users see exactly which fields they passed
and that more is needed to disambiguate.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* chore: fix CI — collapse short ValueError to one line, refresh uv.lock
* `ruff format` on CI (newer version) wants the short `camera=None`
ValueError on a single line.
* `uv.lock` was stale relative to `pyproject.toml`'s `datasets>=4.7.0`
pin (and picked up upstream `s390x` marker fixes for cuda packages).
CI runs `uv sync --locked` which rejected the divergence.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(language): keep base install green — drop processor re-export, gate dataset-extra tests
`lerobot.processor` re-exported `RenderMessagesStep` at the package
level, so importing anything from `lerobot.processor` pulled in
`lerobot.datasets.language` → `lerobot.datasets/__init__.py` →
`require_package("datasets")`, which fails in the Tier 1 base install
that intentionally omits the `[dataset]` extra. The chain bricked
collection for unrelated suites (`tests/policies/pi0_pi05/...`,
`tests/envs/...`, etc.).
* Stop re-exporting `RenderMessagesStep` from `lerobot.processor`. The
only consumer (the test) already imports from the submodule.
Document the deliberate omission in the module docstring.
* Add `pytest.importorskip("datasets", ...)` (and `pandas` where
needed) at the top of the four PR-added tests that exercise the
language stack:
- tests/datasets/test_language.py
- tests/datasets/test_language_render.py
- tests/processor/test_render_messages_processor.py
- tests/utils/test_collate.py
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(language): address review — tools accessor, motion docs, conditional collate
* **`meta.tools` actually reads `info.json["tools"]`.** `DatasetInfo`
had no `tools` field, so `from_dict` silently dropped the key (it
warned about unknown fields then discarded them) and the property
always returned `DEFAULT_TOOLS`. Added `tools: list[dict] | None`
to the dataclass; `to_dict()` drops it when unset so existing
datasets keep a clean `info.json`. Fixed the accessor to read
`self.info.tools` (the previous `.get(...)` would have raised
AttributeError on the dataclass anyway). Added regression tests:
fallback when absent, round-trip from disk, and round-trip
through `DatasetInfo.from_dict` / `to_dict`.
* **`motion` is not view-dependent — fix the docs.** The mdx claimed
rows of style `motion` must carry `camera`, but `VIEW_DEPENDENT_STYLES
= {"vqa", "trace"}` and the validator agrees: motion primitives are
joint/Cartesian-frame, not pixel-space. Updated both call-out
paragraphs in `language_and_recipes.mdx`.
* **Conditional `collate_fn` swap.** Added `meta.has_language_columns`
and gate the `lerobot_collate_fn` swap in `lerobot_train.py` on it,
so non-language datasets keep PyTorch's `default_collate`. Also
added a pass-through test in `test_collate.py` that asserts on a
plain tensor batch the custom collate matches `default_collate`
key-for-key, plus a test for the `None`-sample drop path.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* review: dedupe regex, centralize column names, harden collate, more tests
* **#2 — dedupe `_PLACEHOLDER_RE`.** The same regex was compiled in
`recipe.py` and `language_render.py`. Promote to module-level
`PLACEHOLDER_RE` in `recipe.py` (its primary owner — declares
template syntax) and import from `language_render.py`.
* **#3 — centralize language column names.** `io_utils.py` had
hardcoded `{"language_persistent", "language_events"}` literals at
two sites. Replace with `LANGUAGE_COLUMNS` import so a future column
rename can't silently desync.
* **#4 — defensive collate preserved-keys.** `lerobot_collate_fn`
silently filtered language fields from samples that didn't have
them, which would hand downstream consumers a preserved list
shorter than the tensor batch. Now: if any sample carries a key,
every sample in the batch must carry it; otherwise raise a
`ValueError` so the upstream rendering bug surfaces at the boundary.
* **#5 — `_scalar` rejects non-singleton lists.** Previously a zero-
or multi-element list fell through and triggered confusing
`float([])` errors downstream. Now raises `ValueError` with the
actual length.
* **#6 — refactor `_extract_complementary_data`.** Replace 11 lines
of `key = {... if ... else {}}` plus an 11-line splat dict with a
single `_COMPLEMENTARY_KEYS` tuple iterated once.
* **#7 — document `EXTENDED_STYLES`.** Was an empty `set()` with no
comment. Add a docstring explaining it's an intentional extension
point: downstream modules append project-local styles before
`column_for_style` is called.
* **#9 — `tools.mdx` notes the runtime layer is future work.** The
page referenced `src/lerobot/tools/`, `registry.py`, and
`get_tools(meta)` — none exist in this PR. Added a callout at the
start of "How to add your own tool" plus a note on the
implementations paragraph.
* **#10 — tests for YAML round-trip, malformed rows, blend
validation.** `test_recipe.py` grew from 1 case to 12 covering:
blend-or-messages exclusivity, target-turn requirement, blend
emptiness, weight presence/positivity, nested-blend rejection,
`from_dict` with nested blends, `from_yaml` / `load_recipe`
agreement, top-level non-mapping rejection. Added a malformed-row
test for `_normalize_rows` that asserts non-dict entries raise
`TypeError`.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* review: emitted_at uses 0.1s tolerance; MessageTurn requires stream at construction
* **Float tolerance in `emitted_at` for persistent styles.** The
``_timestamp(row) == t`` exact-equality check silently missed any
caller that derived ``t`` arithmetically (e.g. ``frame_idx / fps``)
even though the parquet timestamp would only differ by ULPs. Added
``EMITTED_AT_TOLERANCE_S = 0.1`` and check ``abs(...) <= tolerance``
instead, with a docstring explaining why exact equality wasn't
enough and why 0.1 s is safe at typical 30–100 Hz control rates.
Test asserts the new behavior at half-window (matches) and
double-window (no match) using the constant so it stays in sync.
* **`MessageTurn.stream` is required at construction.** It was typed
``MessageStream | None = None`` so YAML could omit ``stream:`` and
pass the dataclass invariant — but ``_validate_rendered`` rejected
``None`` streams later, surfacing the error at the first sample
instead of at recipe load. Now ``__post_init__`` raises
``ValueError`` if ``stream`` is ``None``, with the list of valid
streams in the message. The redundant late-stage check in
``_validate_rendered`` is replaced with a one-line comment that
cites the upstream invariant. Test pins the new construction-time
rejection.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* docs(tools): drop follow-up-PR references
Reword the two callouts in `tools.mdx` to describe the runtime layer
in present tense ("not part of the catalog layer shipped today",
"those modules don't yet exist in the tree") instead of pointing at a
specific follow-up PR. Keeps the doc honest about what works now
without coupling it to a particular release order.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* review: address CarolinePascal feedback
- language timestamps: float64 -> float32 to match LeRobotDataset frame
timestamps (Arrow struct + HF feature)
- dataset_metadata: hoist `.language` imports to module top — language.py
has no lerobot imports, so there is no circular-import risk
- dataset_metadata: add a `meta.tools` setter that persists the catalog to
info.json and reloads `meta.info`
- feature_utils: validate the `language` dtype instead of returning "" —
warn (non-fatal) when a non-empty value is written at record time
- centralize the scalar-unwrap helper as `lerobot.utils.utils.unwrap_scalar`,
shared by render_messages_processor and language_render
- docs: move `## Layer 2 — recipe anatomy` ahead of the resolver sections,
which describe recipe bindings rather than dataset layout
- language_render: note in EMITTED_AT_TOLERANCE_S that persistent rows change
on a human-action timescale, not the camera frame rate
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Pepijn