Downloads · 30 days
0
pi0-fast-libero-jax
Fisher-Wang/pi0-fast-libero-jax
pi0-fast-libero-jax is a machine learning model from Fisher-Wang. Use it for the machine learning 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.
Physical Intelligence's official pi0fastlibero checkpoint (gs://openpi-assets/checkpoints/pi0fastlibero, π₀-FAST fine-tuned on the LIBERO benchmark), re-hosted in its original JAX format (params/ + assets/). π₀-FAST i…
Access
Public
Updated Jul 12, 2026
Repo size
10.8 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
Other10.8 GB · 100%
At a glance
- License
- apache-2.0
- Access
- Public
- Created
- Jul 12, 2026
- Updated
- Jul 12, 2026
- SHA
- f1ad17b3
Model card
From the Hugging Face model README
pi0-fast-libero-jax
Physical Intelligence's official pi0_fast_libero checkpoint
(gs://openpi-assets/checkpoints/pi0_fast_libero, π₀-FAST fine-tuned on the
LIBERO benchmark), re-hosted in its original JAX format (params/ +
assets/). π₀-FAST is an autoregressive VLA (FAST action tokenizer); openpi's
PyTorch implementation does not support this architecture, so no PyTorch
conversion exists — serve it with openpi's JAX stack.
Uploaded for RobotTrain subnet validator pipeline testing. Like
pi0-libero-pytorch, this is a genuine LIBERO policy with its own
norm_stats, providing another positive sample for benchmark discrimination
tests.
Load with openpi: create_trained_policy(get_config("pi0_fast_libero"), <this dir>).
- License
- apache-2.0
- Created
- Jul 12, 2026
- Updated
- Jul 12, 2026