skills.rest
qcmuu
qcmuuCOMMUNITY
Explore all skills by this author →
GitHub

fine-tuning-serving-openpi

Fine-tune and serve OpenPI robot policies for manipulation datasets.

Updated May 2, 2026
Search Tags:#fine-tuning#openpi#checkpoint conversion#robot policy#jax training#websocket inference#droid libero aloha
One-click install
npx skills add https://github.com/qcmuu/AI-Research-Skills --skill fine-tuning-serving-openpi-qcmuu
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: fine-tuning-serving-openpi
Source: https://github.com/qcmuu/AI-Research-Skills/tree/main/18-multimodal/openpi
Command: npx skills add https://github.com/qcmuu/AI-Research-Skills --skill fine-tuning-serving-openpi-qcmuu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill solves the problem of getting Physical Intelligence OpenPI (pi0/pi0-fast/pi0.5) to work on your own robot manipulation tasks by providing an end-to-end path from fine-tuning to deployable policy inference.

Core Features & Use Cases

  • JAX fine-tuning workflow: Computes required norm statistics, trains pi0/pi0.5 configs, and produces checkpoints ready for serving.
  • PyTorch training and checkpoint conversion: Converts JAX checkpoints into a PyTorch-compatible format and runs training or inference with converted artifacts.
  • WebSocket policy serving for real-time control: Launches an inference server and provides a client API to send observations and receive actions for environments like ALOHA, DROID, and LIBERO.

Quick Start

Run uv to start the OpenPI policy inference server for DROID using the provided entrypoint and default checkpoint settings.

Frequently Asked Questions about fine-tuning-serving-openpi

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fine-tune OpenPI pi0 models on custom robot manipulation datasets?▼

Fine-tune OpenPI pi0 models by computing required norm statistics before training, running JAX training from supported openpi configs, and producing checkpoints ready for serving.

Can I convert JAX checkpoints to PyTorch for OpenPI policy training?▼

Convert JAX checkpoints to a PyTorch-compatible format to run training or inference with converted artifacts for pi0 robot policies.

How do I deploy a WebSocket inference server for OpenPI robot policies?▼

Deploy a WebSocket policy server for real-time control by launching the inference server and using the client API to send observations and receive actions for ALOHA, DROID, and LIBERO environments.

What norm statistics are required before training OpenPI robot policies?▼

Compute required norm statistics before training OpenPI robot policies to normalize custom manipulation datasets for JAX training and reliable action outputs.

Does OpenPI policy serving support ALOHA, DROID, and LIBERO evaluation environments?▼

OpenPI policy serving supports WebSocket-based policy inference across ALOHA, DROID, and LIBERO evaluation environments for real-time robot control.

Do I need uv to manage dependencies for OpenPI fine-tuning and serving?▼

OpenPI fine-tuning and serving requires uv-managed dependencies to launch the inference server and run the JAX training workflows.

Related Skills

Explore More Software Engineering Skills→
tadod12tadod12

fine-tuning-serving-openpi

Fine-tune and serve OpenPI robot policies across ALOHA, DROID, and LIBERO environments.

Community
Advanced
OmidZamaniOmidZamani
120

dspy-finetune-bootstrap

Fine-tune a DSPy program into deployable model weights via BootstrapFinetune.

Community
Advanced
handsomelong922handsomelong922

fine-tuning-serving-openpi

Fine-tune and serve OpenPI models for robot policy inference.

Community
Advanced
KappTech88KappTech88

fine-tuning-serving-openpi

Fine-tune and serve OpenPI robot policy models for real-time control.

Community
Advanced