OpenPI Fine Tuning and Serving
Skill Verified ActiveFine-tune and serve Physical Intelligence OpenPI models (pi0, pi0-fast, pi0.5) using JAX or PyTorch backends for robot policy inference across ALOHA, DROID, and LIBERO environments. Use when adapting pi0 models to custom datasets, converting JAX checkpoints to PyTorch, running policy inference servers, or debugging norm stats and GPU memory issues.
To enable researchers and engineers to adapt, train, and deploy OpenPI models for robot policy inference, streamlining complex ML workflows.
Features
- Fine-tune pi0, pi0-fast, pi0.5 models
- Support JAX and PyTorch backends
- Convert JAX checkpoints to PyTorch
- Serve policies via WebSocket API
- Automated environment setup and dependency management
Use Cases
- Adapting pi0 models to custom datasets
- Converting JAX checkpoints to PyTorch
- Running policy inference servers for robot control
- Debugging norm stats and GPU memory issues during training
Non-Goals
- Training or serving of models other than OpenPI variants
- General-purpose machine learning framework utilities
- Automated dataset curation or generation
Workflow
- Set up environment (clone repo, sync dependencies)
- Select and configure training parameters
- Compute normalization statistics
- Launch training (JAX or PyTorch)
- Convert JAX checkpoints to PyTorch (if needed)
- Serve trained policies
- Integrate client into robot/simulation code
Practices
- Model Fine-Tuning
- Policy Serving
- JAX/PyTorch Development
- Checkpoint Management
Prerequisites
- JAX >= 0.4.30
- PyTorch >= 2.1.0
- Transformers >= 4.53.2
- uv >= 0.4.0
Installation
First, add the marketplace
/plugin marketplace add Orchestra-Research/AI-Research-SKILLs/plugin install AI-Research-SKILLs@ai-research-skillsQuality Score
VerifiedTrust Signals
Similar Extensions
Implementing Llms Litgpt
100Implements and trains LLMs using Lightning AI's LitGPT with 20+ pretrained architectures (Llama, Gemma, Phi, Qwen, Mistral). Use when need clean model implementations, educational understanding of architectures, or production fine-tuning with LoRA/QLoRA. Single-file implementations, no abstraction layers.
OpenVLA OFT Fine Tuning and Evaluation
98Fine-tunes and evaluates OpenVLA-OFT and OpenVLA-OFT+ policies for robot action generation with continuous action heads, LoRA adaptation, and FiLM conditioning on LIBERO simulation and ALOHA real-world setups. Use when reproducing OpenVLA-OFT paper results, training custom VLA action heads (L1 or diffusion), deploying server-client inference for ALOHA, or debugging normalization, LoRA merge, and cross-GPU issues.
Unsloth
100Expert guidance for fast fine-tuning with Unsloth - 2-5x faster training, 50-80% less memory, LoRA/QLoRA optimization
PyTorch Lightning
100Deep learning framework (PyTorch Lightning). Organize PyTorch code into LightningModules, configure Trainers for multi-GPU/TPU, implement data pipelines, callbacks, logging (W&B, TensorBoard), distributed training (DDP, FSDP, DeepSpeed), for scalable neural network training.
Peft Fine Tuning
99Parameter-efficient fine-tuning for LLMs using LoRA, QLoRA, and 25+ methods. Use when fine-tuning large models (7B-70B) with limited GPU memory, when you need to train <1% of parameters with minimal accuracy loss, or for multi-adapter serving. HuggingFace's official library integrated with transformers ecosystem.
Huggingface Llm Trainer
99Train or fine-tune language and vision models using TRL (Transformer Reinforcement Learning) or Unsloth with Hugging Face Jobs infrastructure. Covers SFT, DPO, GRPO and reward modeling training methods, plus GGUF conversion for local deployment. Includes guidance on the TRL Jobs package, UV scripts with PEP 723 format, dataset preparation and validation, hardware selection, cost estimation, Trackio monitoring, Hub authentication, model selection/leaderboards and model persistence. Use for tasks involving cloud GPU training, GGUF conversion, or when users mention training on Hugging Face Jobs without local GPU setup.