mirror of
https://github.com/huggingface/lerobot.git
synced 2026-05-19 18:49:52 +00:00
Pepijn Kooijmans
fd18beb3a1
- name the three modules everywhere (plan / interjections / vqa) instead of module_1/2/3 — config classes, config fields, executor params, staging keys and phase names now carry the module name - rename examples/annotation -> examples/annotations; add the Apache header to run_hf_job.py - drop the unused GeneralVqaModule._generate_one - remove "PR 1" references from comments/docstrings - frames.py: rely on the always-defined LeRobotDatasetMetadata.camera_keys - executor.py: read/write meta/info.json via load_info / write_info - reader.py: load meta/tasks.parquet via io_utils.load_tasks - make --push_to_hub a bool; push the annotated dataset back to --repo_id - move the on-disk test dataset builder into tests/fixtures (build_annotation_dataset); run_e2e_smoke reuses it - clarify in the docs that the vqa module grounds each pair on a single frame (K = per-tick anchor count) - hoist stdlib dynamic imports to module scope Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
183 lines
8.5 KiB
Plaintext
183 lines
8.5 KiB
Plaintext
# Annotation Pipeline
|
||
|
||
`lerobot-annotate` populates the two language columns introduced by the
|
||
[Language Columns and Recipes](./language_and_recipes) page —
|
||
`language_persistent` and `language_events` — directly into
|
||
`data/chunk-*/file-*.parquet`.
|
||
|
||
## What the pipeline produces
|
||
|
||
Three modules write into a per-episode staging tree, then a single writer
|
||
rewrites the data shards in place:
|
||
|
||
| Style / atom | Column | Module |
|
||
| ------------------------------------------- | --------------------- | -------------- |
|
||
| `subtask` (Pi0.7-style "how, not what") | `language_persistent` | `plan` |
|
||
| `plan` (initial + refresh on interjection) | `language_persistent` | `plan` |
|
||
| `memory` (MEM-style compression) | `language_persistent` | `plan` |
|
||
| `task_aug` (rephrasings of canonical task) | `language_persistent` | `plan` |
|
||
| `interjection` | `language_events` | `interjections`|
|
||
| speech tool-call atom (`style=null`, `say`) | `language_events` | `interjections`|
|
||
| `vqa` (user / assistant pair) | `language_events` | `vqa` |
|
||
|
||
The writer does **not** add a `tools` column to the parquet — the tool
|
||
catalog lives at `meta/info.json["tools"]` instead (see
|
||
[Tools](./tools)). After every annotation run the pipeline ensures the
|
||
canonical `say` schema is present in that list, preserving any tools the
|
||
user pre-declared.
|
||
|
||
If you want to declare additional tools for a dataset before annotation
|
||
runs, edit `meta/info.json["tools"]` directly — the pipeline preserves
|
||
anything already there. Implementations of those tools live under
|
||
`src/lerobot/tools/`; one file per tool, registered via
|
||
`TOOL_REGISTRY`. See the [Tools](./tools) doc for the authoring guide.
|
||
|
||
## Running locally
|
||
|
||
Install the extra and invoke the console script. Episode-level
|
||
concurrency comes from `--executor.episode_parallelism` (default 16);
|
||
that is the only knob the in-process executor exposes.
|
||
|
||
```bash
|
||
uv sync --extra annotations
|
||
uv run lerobot-annotate \
|
||
--root=/path/to/dataset \
|
||
--vlm.model_id=Qwen/Qwen2.5-VL-7B-Instruct
|
||
```
|
||
|
||
The pipeline attaches actual camera footage to every `plan` /
|
||
`interjections` / `vqa` prompt by default, decoded from the dataset's
|
||
first `observation.images.*` stream. Override with
|
||
`--vlm.camera_key=observation.images.<name>` to pin a specific
|
||
viewpoint. Datasets with no video tracks fall back to text-only prompts
|
||
automatically.
|
||
|
||
**The `plan` module sees the whole episode as one video block.** Subtask
|
||
decomposition gets a `{"type":"video", "video":[<frames>]}` block
|
||
covering the entire demonstration; Qwen-VL pools temporally on its own
|
||
and decides where to cut. There is no keyframe stride or count knob —
|
||
`--plan.max_video_frames` (default 128) only caps the frames packed
|
||
into the video block as a model-capacity bound. The `interjections`
|
||
module attaches a short window of frames straddling the interjection
|
||
timestamp. The `vqa` module grounds each VQA pair on a single frame —
|
||
its `--vqa.K` knob sets how many consecutive frames each emission tick
|
||
anchors, and every anchored frame gets its own VQA pair on that one
|
||
frame (there is no per-pair frame window).
|
||
|
||
## Running on Hugging Face Jobs
|
||
|
||
Distributed annotation is delegated to
|
||
[Hugging Face Jobs](https://huggingface.co/docs/hub/en/jobs). The repo
|
||
ships a launcher script you copy and edit for your dataset:
|
||
|
||
```bash
|
||
HF_TOKEN=hf_... uv run python examples/annotations/run_hf_job.py
|
||
```
|
||
|
||
[`examples/annotations/run_hf_job.py`](https://github.com/huggingface/lerobot/blob/main/examples/annotations/run_hf_job.py)
|
||
spawns one `h200x2` job that:
|
||
|
||
1. installs the branch under test plus the annotation extras,
|
||
2. boots two vllm servers (one per GPU) for the chosen model,
|
||
3. runs the `plan` / `interjections` / `vqa` modules across the dataset
|
||
via `lerobot-annotate`,
|
||
4. uploads the annotated dataset to `--push_to_hub`.
|
||
|
||
To target a different dataset, model, or hub repo, edit the `CMD` block
|
||
inside the script — every flag in there maps directly onto a CLI flag of
|
||
`lerobot-annotate` (see `lerobot-annotate --help` for the full list).
|
||
|
||
## Style-to-recipe consumer mapping
|
||
|
||
The pipeline's outputs are designed to be consumed by recipes (see
|
||
[Language Columns and Recipes](./language_and_recipes)) — typically:
|
||
|
||
- low-level / high-level / memory-update branches consume
|
||
`subtask`/`plan`/`memory` from `language_persistent`.
|
||
- An interjection-response branch consumes `interjection` events plus
|
||
the paired speech atom (merged into one assistant target turn via
|
||
`tool_calls_from`) and the same-timestamp `plan` refresh.
|
||
- A VQA branch consumes the `(vqa, user)` and `(vqa, assistant)` pairs
|
||
from `language_events`.
|
||
|
||
## Why the design splits state from events
|
||
|
||
Two things drive the scope:
|
||
|
||
1. **Persistent state vs exact-event split.** Persistent rows
|
||
(`subtask`, `plan`, `memory`) broadcast per episode and answer "what
|
||
state is in force at this frame?". Event rows (`interjection`, `vqa`,
|
||
speech) only appear on the exact frame whose timestamp matches the
|
||
emission. The pipeline writes timestamps taken straight from the
|
||
source parquet — no floating-point recomputation.
|
||
2. **One Qwen-VL pass.** All three modules share a single VLM client
|
||
(vLLM if available, transformers fallback) so the cost is one model
|
||
load per dataset, not three.
|
||
|
||
## Module independence and staged reruns
|
||
|
||
Each module writes its raw output to
|
||
`<root>/.annotate_staging/episode_{N:06d}/<module>.jsonl`. That makes
|
||
prompt iteration cheap — re-running one module overwrites only its own
|
||
JSONL file before the writer composes the final parquet. Modules can be
|
||
disabled via `--plan.enabled=false` (and likewise `--interjections.enabled`
|
||
/ `--vqa.enabled`) to
|
||
test them in isolation.
|
||
|
||
## Validation/report checks before final write
|
||
|
||
Before the writer runs, `StagingValidator` checks:
|
||
|
||
- exact frame-timestamp alignment for every event row;
|
||
- no orphan speech / interjection pairs;
|
||
- `plan` is refreshed at every interjection timestamp;
|
||
- `memory` rows fall on subtask boundaries (warning, not error);
|
||
- VQA assistant `content` parses as JSON in one of the
|
||
bbox / keypoint / count / attribute / spatial shapes;
|
||
- every row routes to the column dictated by `column_for_style(style)`.
|
||
|
||
Errors abort the writer (`--skip_validation=true` overrides for debugging).
|
||
|
||
## Paper inspirations per module
|
||
|
||
- **`plan` module — subtasks.** Hi Robot ([Shi 2025](https://arxiv.org/abs/2502.19417))
|
||
atom granularity ("pick up one piece of lettuce", "place bowl to box");
|
||
Pi0.7 ([Physical Intelligence 2025](https://pi.website/pi07)) "how, not
|
||
what" detail.
|
||
- **`plan` module — memory.** MEM ([Torne 2026](https://arxiv.org/abs/2603.03596))
|
||
compression directive: keep only minimal relevant information; functional
|
||
outcomes preserved, specific attributes dropped.
|
||
- **`interjections` module.** Hi Robot scenario taxonomy: negative task,
|
||
situated correction, specific constraint, preference. Speech is a
|
||
tool-call-only atom (`tool_calls=[{type:function, function:{name:"say",
|
||
arguments:{text:...}}}]`).
|
||
- **`vqa` module.** ECoT ([Zawalski 2024](https://arxiv.org/abs/2407.08693))
|
||
grounded features (bounding boxes in pixel `[x_min, y_min, x_max, y_max]`,
|
||
keypoints) and Steerable VLA Policies ([Zhao 2025](https://arxiv.org/abs/2509.07626))
|
||
multi-abstraction grounding. Pi0.7 also grounds answers across
|
||
multiple abstraction levels.
|
||
|
||
Future maintainers should adjust the prompt templates in
|
||
`src/lerobot/annotations/steerable_pipeline/prompts/` against these
|
||
references rather than rewriting from scratch.
|
||
|
||
## Compute and list-size estimates
|
||
|
||
Per episode, the pipeline issues O(`max_steps`) `plan`-module calls,
|
||
O(`max_interjections_per_episode`) `interjections`-module calls, and
|
||
O(`vqa_emission_hz × episode_seconds`) `vqa`-module calls. With defaults
|
||
(8 subtasks, 1 interjection, 1 Hz × 3 pairs) and 30-second episodes, that
|
||
is ~50 VLM calls per episode. `language_persistent` per episode is ~10s of
|
||
KB at most (parquet dictionary-encodes one entry per episode);
|
||
`language_events` is empty on most frames and is bounded by the number of
|
||
emissions, not `num_frames × num_emissions`.
|
||
|
||
## Reproducibility via seed and prompt hashes
|
||
|
||
`--seed` (default 1729) feeds the per-episode RNGs that select interjection
|
||
timestamps and VQA question types. Combined with the deterministic prompt
|
||
templates checked into `prompts/`, two runs at the same seed against the
|
||
same dataset and the same model checkpoint produce byte-identical staging
|
||
artifacts. Prompt edits are recorded by file hash; future tooling can pin
|
||
expected `(seed, prompt_hash)` pairs into the dataset card.
|