metadata
license: apache-2.0
tags:
- openpi
- pi0.5
- robotics
- vla
- piper
- cube-stacking
pi05_piper_stacking_v3 - step 7500
pi0.5 (action_horizon 50) full fine-tune of pi05_base on axiboai/piper_stacking_v3 (155 eps = v2 + ~44 corrections).
Latest checkpoint pushed: step 7500 / 10000. Inference artifacts only (params + assets/norm_stats; no train_state).
Prompt: stack red cube on blue cube
import pathlib
from huggingface_hub import snapshot_download
from openpi.training import config as _config
from openpi.policies import policy_config
ckpt = pathlib.Path(snapshot_download("axiboai/pi05_piper_stacking_v3"))
policy = policy_config.create_trained_policy(_config.get_config("pi05_piper_stacking_v3"), ckpt)