Pepijn
f763f85213
feat: language annotation pipeline (PR 2/3)
Adds the steerable annotation pipeline (`lerobot-annotate`) that populates
the `language_persistent` and `language_events` columns introduced in
PR 1 directly into `data/chunk-*/file-*.parquet`. No flavor namespace,
no sidecar tree.
Modules produced:
- Module 1 (plan_subtasks_memory): Pi0.7-style subtasks, plan (init +
refresh on interjection), MEM-style memory at subtask boundaries.
- Module 2 (interjections_and_speech): t=0 speech-only acknowledgement,
mid-episode paired interjection + speech tool-call atom.
- Module 3 (general_vqa): bbox/keypoint/count/attribute/spatial pairs at
configurable cadence with one-retry JSON validation.
Writer enforces: per-episode persistent identity, exact-frame event
timestamps, column routing per `column_for_style`, dataset-level `tools`
column with the `say` schema, drops legacy `subtask_index`. Validator
runs against staged JSONL artifacts before the writer rewrites parquet.
Adds `lerobot-annotate` console script, `annotations` extra (datatrove +
optional vllm), `make annotation-e2e` opt-in smoke target, and
`docs/source/annotation_pipeline.mdx`.
Branched from PR 1 (`feat/language-columns`).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-30 18:48:33 +02:00
..
2026-04-30 18:48:33 +02:00
2026-02-23 13:57:43 +01:00
2026-04-01 12:59:12 +02:00
2026-04-12 20:03:04 +02:00
2026-04-30 18:48:33 +02:00
2026-04-12 20:03:04 +02:00
2026-04-12 20:03:04 +02:00
2026-04-12 20:03:04 +02:00
2026-04-12 20:03:04 +02:00
2025-06-05 17:48:43 +02:00
2026-01-26 17:53:25 +01:00
2025-09-18 15:25:26 +02:00
2026-04-12 20:03:04 +02:00
2026-04-12 20:03:04 +02:00
2026-04-12 20:03:04 +02:00
2026-04-12 20:03:04 +02:00
2026-04-12 20:03:04 +02:00
2025-08-28 11:18:54 +02:00
2026-04-05 19:09:15 +02:00
2026-04-02 19:53:59 +02:00
2025-09-23 16:32:34 +02:00
2026-04-14 11:28:49 +02:00
2026-02-23 13:57:43 +01:00
2026-04-16 17:53:35 +02:00
2025-09-18 15:25:26 +02:00
2025-07-17 14:30:20 +02:00
2026-04-12 20:03:04 +02:00
2026-01-08 13:04:30 +01:00
2026-04-12 20:03:04 +02:00
2025-09-24 16:17:39 +02:00
2026-04-30 10:48:17 +02:00
2026-03-04 15:08:49 +01:00
2026-04-12 20:03:04 +02:00
2026-04-20 21:07:21 +02:00
2026-04-03 14:44:53 +02:00
2026-04-09 10:29:20 +02:00
2026-04-12 20:03:04 +02:00
2026-04-05 21:23:13 +02:00
2025-07-17 14:30:20 +02:00
2026-01-25 22:29:37 +01:00
2026-01-28 17:25:57 +01:00
2026-01-05 08:51:26 +01:00
2026-04-12 20:03:04 +02:00
2026-04-12 20:03:04 +02:00
2026-03-06 15:52:54 +01:00
2026-04-12 20:03:04 +02:00
2025-08-01 17:39:39 +02:00
2025-08-01 17:39:39 +02:00
2025-10-23 13:50:30 +02:00
2026-03-28 00:41:26 +01:00
2026-04-12 20:03:04 +02:00
2026-04-05 19:25:38 +02:00
2026-04-05 19:25:38 +02:00
2026-04-05 19:25:38 +02:00
2025-08-01 17:39:39 +02:00
2025-08-01 17:39:39 +02:00
2025-08-01 17:39:39 +02:00
2026-01-06 15:13:35 +01:00
2025-09-15 09:53:30 +02:00
2026-01-08 14:45:07 +01:00
2026-02-23 13:57:43 +01:00
2026-03-23 13:57:53 -07:00
2026-04-20 17:10:53 +02:00
2026-04-20 19:12:15 +02:00
2026-04-20 20:21:27 +02:00
2026-04-20 17:46:39 +02:00
2026-04-12 20:03:04 +02:00
2026-02-18 22:46:12 +01:00
2026-02-23 13:57:43 +01:00
2026-01-08 13:04:30 +01:00
2026-03-31 12:16:34 +02:00
2026-02-23 13:57:43 +01:00
2026-04-30 18:44:58 +02:00
2025-12-05 13:32:58 +01:00
2026-03-09 18:47:12 +01:00
2026-02-17 20:09:42 +01:00
2026-04-21 17:54:11 +02:00
2026-02-18 22:46:12 +01:00
2026-04-12 20:03:04 +02:00