Downloads · 30 days
0
pi05-pick_and_place_N_v1-fp
sm-park/pi05-pick_and_place_N_v1-fp
pi05-pick_and_place_N_v1-fp is a robotics model from sm-park. Use it for the robotics task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Full-precision (bf16) PyTorch full fine-tune of pi05base on the sm-park/pickandplaceNv1 LeRobot dataset (single-arm "Aloha solo" pick-and-place, 9-dim state/action: 8 joints + 1 gripper).
Access
Public
Updated Aug 30, 2026
Parameters
3.6B
7.5 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors7.5 GB · 100%
Parameter types
How the weights are stored.
BF163.5B · 97%
Model card
From the Hugging Face model README
pi0.5 full fine-tune — sm-park/pick_and_place_N_v1
Full-precision (bf16) PyTorch full fine-tune of pi05_base on the
sm-park/pick_and_place_N_v1
LeRobot dataset (single-arm "Aloha solo" pick-and-place, 9-dim state/action: 8 joints + 1 gripper).
- Base model:
gs://openpi-assets/checkpoints/pi05_base/params(openpi) - Training config:
pi05_pick_and_place_N_v1insrc/openpi/training/config.py - 200,000 steps, batch size 32, PyTorch (
scripts/train_pytorch.py) - Final training loss: 0.0005
Contents
model.safetensors— PyTorch model weights (PI0Pytorch)assets/sm-park/pick_and_place_N_v1/norm_stats.json— state/action normalization stats
Usage
Load with openpi's PyTorch model loading utilities
(openpi.models_pytorch.pi0_pytorch.PI0Pytorch), using the pi05_pick_and_place_N_v1 config for
architecture (action_dim=32, action_horizon=50, pi05=True) and the included norm stats.
- Task
- Robotics
- License
- apache-2.0
- Created
- Aug 30, 2026
- Updated
- Aug 30, 2026