lerobot

mirror of https://github.com/huggingface/lerobot.git synced 2026-05-15 16:49:55 +00:00

Files

T

Pepijn 12cce8f2cc fix(smolvla2): align flow_loss_weight default with Pi 0.5 paper's α=10

Pi 0.5 paper §IV.D Eq. (1) sets the loss balance to α=10 between text
CE and flow MSE: actions are the primary output and the flow head
should dominate the gradient signal. SmolVLA2 was defaulting both
weights to 1.0, which inverts that — text CE (~0.5-2.0 nats) ends up
larger than flow MSE (~0.1-1.0), so the action expert gets less
gradient than the LM head despite being the primary task.

Match the paper's split: text_loss_weight=1.0, flow_loss_weight=10.0.
Same as ``pi052`` (the new full reproduction policy).

Also pin the values explicitly in the SLURM launcher so the choice is
visible and overridable per-run rather than buried in the config
default.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

2026-05-13 11:02:17 +02:00

annotation

chore(annotate): throttle Module 3 + executor parallelism to fix vLLM stall

2026-05-05 15:07:18 +02:00

backward_compatibility

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

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

dataset

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

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

hil

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