admin/lerobot
1
0
Fork 0
mirror of https://github.com/huggingface/lerobot.git synced 2026-07-20 08:21:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor(pi052): self-contained policy; revert pi0/pi05 to upstream main

Browse Source 
The smolvla branch had modified the shared pi0/pi05 modeling + pi05 config to
support pi052 (SDPA attention, layernorm/lm_head handling, optimizer
foreach/fused/lm_head_lr_scale, embedding scaling). Decouple pi052 instead:

- Vendor the PI0.5 backbone (PaliGemmaWithExpertModel, PI05Pytorch, helpers)
  into pi052/pi05_backbone.py (verbatim copy, no PI05Policy).
- Flatten PI052Policy to subclass PreTrainedPolicy directly (no longer
  PI05Policy); inline the needed PI05Policy methods.
- Restore optimizer_foreach/fused + get_optimizer_preset on PI052Config.
- Revert pi0, pi0_fast, pi05 modeling and configuration_pi05 to origin/main
  (byte-identical), so the shared policies carry no smolvla modifications.

Behavior verified bit-exact on pepijn223/pi052_robocasa_full: embed_language_
tokens, predict_action_chunk, and the fused flow+text+FAST training loss are
identical before/after (max_abs_diff=0). pi052 tests pass (pre-existing
stale-name collection errors unchanged).

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
pepijn223
2026-06-04 19:59:27 +02:00
parent 8292548f0d
commit a594ad7969
8 changed files with 1484 additions and 204 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/policies/pi052/test_pi052_sdpa_attention.py
+1 -1
View File
@@ -33,7 +33,7 @@ pytest.importorskip("transformers")
from transformers.models.gemma import modeling_gemma # noqa: E402
from lerobot.policies.pi05.modeling_pi05 import ( # noqa: E402
from lerobot.policies.pi052.pi05_backbone import ( # noqa: E402
make_att_2d_masks,
sdpa_attention_forward,
)