All issues

Search results

[Evaluation] Full eval is very slow to reproduce — expected? ways to speed up?

Status: Open.

#25 In mli0603/openpi-comet;

· liwq229 opened on Jun 15, 2026

Reproduction Issue: Extremely low Q-score (~0.03 / 3%) on BEHAVIOR-1K 50 tasks with pi05-b1kpt50-cs32 despite correct setup; latest main branch causes multiple import/Hydra errors

Status: Open.

#24 In mli0603/openpi-comet;

· riyuu2197 opened on Apr 14, 2026

norm_stats计算问题

Status: Open.

#23 In mli0603/openpi-comet;

· Christopher6488 opened on Mar 18, 2026

the Repeatability Experiment of Policy Evaluation

Status: Open.

#22 In mli0603/openpi-comet;

· TTZYTttzyt opened on Mar 11, 2026

delta joint make 0 success rate?

Status: Open.

#21 In mli0603/openpi-comet;

· wqingzex opened on Mar 2, 2026

Question regarding the training recipe for reproducing public checkpoints

Status: Open.

#20 In mli0603/openpi-comet;

· wjdgns0914 opened on Feb 25, 2026

Questions about evaluation.

Status: Open.

#19 In mli0603/openpi-comet;

· hyn1426536274 opened on Jan 20, 2026

Aligning results and Questions about latest pretrained models

Status: Open.

#17 In mli0603/openpi-comet;

· YushenZuo opened on Jan 15, 2026

Questions while reproducing the results

Status: Open.

#16 In mli0603/openpi-comet;

· Garsonjw opened on Jan 8, 2026

consult about sft

Status: Open.

#15 In mli0603/openpi-comet;

· xiwangxi251 opened on Jan 7, 2026

consult about pretrain

Status: Open.

#14 In mli0603/openpi-comet;

· mafangniu opened on Jan 7, 2026

Inquiry Regarding the Usage and Insights of fine_grained_level Setting in VLA Training and Evaluation

Status: Open.

#5 In mli0603/openpi-comet;

· JarvisUSTC opened on Dec 11, 2025