pepijn 4940281120 feat(streaming): random-episode admission via reshard() + multi-input-shard shuffle ...

Reshard parquet per row group (1 shard == 1 row group == 1 episode) and feed the
episode-pool shuffle with max_buffer_input_shards so the pool is a uniform random
sample of the corpus, independent of episodes-per-file. Add validate_row_groups
guardrails (collapsed-row-group + distributed divisibility), require datasets>=5.0.0,
make the test fixture write one row group per episode, and plumb max_buffer_input_shards
through the dataloading benchmark.

Co-authored-by: Cursor <cursoragent@cursor.com>

2026-06-15 13:33:27 +00:00

..

backward_compatibility

feat(dependencies): minimal default tag install (#3362)

2026-04-12 20:03:04 +02:00

dataset

refactor: support custom progress parquet overlays (#3640)

2026-05-21 14:32:10 +02:00

lekiwi

feat(rollout): decouple policy deployment from data recording with new lerobot-rollout CLI (#3413)

2026-04-28 00:57:35 +02:00

notebooks

docs(quickstart): adding rollout (#3598)

2026-05-14 12:32:39 +02:00

omx

docs(omx): adding some examples and scripts (#3566)

2026-05-11 15:36:32 +02:00

phone_to_so100

feat(rollout): decouple policy deployment from data recording with new lerobot-rollout CLI (#3413)

2026-04-28 00:57:35 +02:00

port_datasets

feat(dependencies): minimal default tag install (#3362)

2026-04-12 20:03:04 +02:00

rtc

feat(rollout): decouple policy deployment from data recording with new lerobot-rollout CLI (#3413)

2026-04-28 00:57:35 +02:00

scaling

feat(streaming): random-episode admission via reshard() + multi-input-shard shuffle

2026-06-15 13:33:27 +00:00

so100_to_so100_EE

feat(rollout): decouple policy deployment from data recording with new lerobot-rollout CLI (#3413)

2026-04-28 00:57:35 +02:00

training

feat(dependencies): minimal default tag install (#3362)

2026-04-12 20:03:04 +02:00

tutorial

RL stack refactoring (#3075)

2026-05-12 15:49:54 +02:00