AI Research Skills Library is a collection of reusable instructions that guide AI agents through research and machine-learning engineering tasks, from finding ideas and writing papers to training, evaluation, and deployment. It is for configuring agents such as Claude Code, Codex, and Gemini to perform research workflows.
Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/Orchestra-Research/AI-Research-SKILLsnpx agentmods add skills/orchestra-research/ai-research-skills/openpiWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/orchestra-research/ai-research-skills/openpi)<a href="https://agentmods.dev/skills/orchestra-research/ai-research-skills/openpi"><img src="https://agentmods.dev/badge/skills/orchestra-research/ai-research-skills/openpi/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/orchestra-research/ai-research-skills/openpi"><img src="https://agentmods.dev/badge/skills/orchestra-research/ai-research-skills/openpi.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00087 | $0.03724 |
| Opus 5 | $0.00044 | $0.01862 |
| Sonnet 5 | $0.00017 | $0.00745 |
| Haiku 4.5 | $0.00009 | $0.00372 |
Grade A, and why
fine-tuning-serving-openpi scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 481 lines — stays where its author put it; the contents beside it link to each section on GitHub.
OpenPI Fine-Tuning and Serving
End-to-end workflows for fine-tuning and serving Physical Intelligence's OpenPI models (pi0, pi0-fast, pi0.5) on robot manipulation tasks from the public openpi repository. Covers blank-machine setup, JAX training, PyTorch training, checkpoint conversion, and policy inference serving.
Quick start
Clone the public repo, install the workspace, then serve a pretrained policy:
git clone --recurse-submodules https://github.com/Physical-Intelligence/openpi.git
cd openpi
GIT_LFS_SKIP_SMUDGE=1 uv sync
GIT_LFS_SKIP_SMUDGE=1 uv pip install -e .
uv run scripts/serve_policy.py --env DROID
from openpi_client import websocket_client_policy
client = websocket_client_policy.WebsocketClientPolicy(host="localhost", port=8000)
result = client.infer(observation)
actions = result["actions"] # numpy array of shape (chunk_size, action_dim)
Core concepts
Model family: OpenPI implements three model variants from Physical Intelligence:
| Model | Architecture | Speed | Quality | Typical use |
|---|---|---|---|---|
| pi0 | Flow-matching VLA | Baseline | Highest | Research, complex tasks |
| pi0-fast | Autoregressive action tokens | 2-5x faster | Good | Real-time control |
| pi0.5 | pi0 + improved vision encoder | Baseline | Best | Latest default |
Key design choices:
- Dual backend: JAX (primary, official training) and PyTorch (community, deployment-friendly)
- Config-driven: All training/serving parameters defined in
src/openpi/training/config.py - Norm stats: Every config requires precomputed normalization statistics before training
- WebSocket serving: Policy servers expose a WebSocket API for low-latency inference
Training loop invariant: After every config or dataset change, always re-run this cycle:
- Compute norm stats → 2. Train → 3. Serve checkpoint → 4. Validate inference
Compute requirements
| Task | GPU | VRAM | Notes |
|---|---|---|---|
| Serve pi0.5 (inference) | 1x A100/H100 | ~24 GB | Single GPU sufficient |
| Fine-tune pi0.5 (JAX) | 1x A100 80GB | ~60 GB | Use fsdp_devices for multi-GPU |
| Fine-tune pi0 (JAX) | 1x A100 80GB | ~40 GB | Smaller model footprint |
| Fine-tune (PyTorch DDP) | 1-8x A100 | ~40 GB/GPU | torchrun launcher |
| Compute norm stats | CPU or 1x GPU | ~8 GB | Fast, can run on login node |
What ships with it
5 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 481 lines · 87 tokens per session scan A 60af709a253b
fine-tuning-serving-openpi is a skill published in the GitHub repository Orchestra-Research/AI-Research-SKILLs (12,508 stars, last pushed 2mo ago), licensed MIT. It adds 87 tokens to every session and 3,724 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
research-cosmos-policy
Evaluates NVIDIA Cosmos Policy on LIBERO and RoboCasa simulation environments. Use when setting up cosmos-policy for robot manipulation evaluation, running headless GPU evaluations with EGL renderi...
geometry
Pure-math 3D geometry toolbox — back-project masks and depth to point clouds, DBSCAN-filter noise, fit oriented bounding boxes, derive top-down/front grasp poses, and reconstruct collision worlds from RGB-D frames. Use when a workflow needs perception geometry or planner inputs computed on CPU with no model weights.
molmoact-libero
Run the MolmoAct LIBERO checkpoint (allenai/MolmoAct-7B-D-LIBERO-0812) as a closed-loop VLA policy for the dexterous pick-and-place segment of a task. Drives a Franka Panda in the LIBERO/robosuite OSCPOSE action space from agentview + wrist cameras, served behind a vLLM-style script speaking the openpi websocket…
huggingface-transformers
Use Hugging Face Transformers pipelines, tokenizers, and AutoModel interfaces for inference and fine-tuning workflows.
ai-for-science-tf-to-pytorch
A general guide for rewriting TensorFlow or Keras machine-learning models in PyTorch. It covers matching layers, converting weights, checking values step by step, and comparing complete model results, including for Ascend NPU environments.
ai-for-science-ankh-ascend-npu-skill
A migration guide for running Ankh protein language models on Huawei Ascend NPU hardware. It covers environment checks, code changes, model-weight loading, validation scripts, and recording the setup.