ai-experiments
Discover scientific datasets, train models, optimize GPU workloads, iterate benchmarks, and record results for AI4Science, LLM, ML, and systems research.
Before you start
Choose your system and agent, then run the command in your terminal. After installation, send the prompts below to your agent.
curl -sSL https://paperskills.com/scripts/paperskills-install.sh | bash -s -- \
--tool codex \
--skills sa.hugging-science,sa.pytorch-lightning,sa.optimize-for-gpu,sa.arbor,air.ml-paper-writing \
--registry https://paperskills.com/api/registryStep 1 · sa.hugging-science
Task for your agent
Use hugging-science to find datasets, pretrained models, spaces, and baseline papers for this scientific ML task, with licensing, I/O, and reproducibility notes.Expected outputs
Step 2 · sa.pytorch-lightning
Task for your agent
Use pytorch-lightning to organize the training code into a configurable, logged, multi-GPU-ready experiment framework with required callbacks and log fields.Expected outputs
Step 3 · sa.arbor
Task for your agent
Use arbor to plan iterative experiment search: hypothesis tree, candidate changes, dev/test isolation, stopping rules, and result review table.Expected outputs
Use when the user is doing AI/ML work in a scientific domain such as biology, chemistry, physics, astronomy, climate, genomics, materials, medicine, ecology, energy, engineering, math, drug discovery, protein design, weather modeling, theorem proving, single-cell, or PDE solving. Hugging Science is a curated catalog of scientific datasets, models, blog posts, and interactive Spaces. This skill helps discover and use resources via `datasets`, `transformers`, the HF Inference API, `gradio_client`, and methodology citations.
View skillDeep learning framework (PyTorch Lightning / lightning package). Organize PyTorch code into LightningModules, configure Trainers for multi-GPU/TPU, implement data pipelines, callbacks, logging (W&B, TensorBoard, MLflow), distributed training (DDP, FSDP, DeepSpeed), for scalable neural network training.
View skillGPU-accelerate Python code using CuPy, Numba CUDA, Warp, cuDF, cuML, cuGraph, KvikIO, cuCIM, cuxfilter, cuVS, cuSpatial, and RAFT. Use whenever the user mentions GPU/CUDA/NVIDIA acceleration, or wants to speed up NumPy, pandas, scikit-learn, scikit-image, NetworkX, GeoPandas, or Faiss workloads. Covers physics simulation, differentiable rendering, mesh ray casting, particle systems (DEM/SPH/fluids), vector/similarity search, GPUDirect Storage file IO, interactive dashboards, geospatial analysis, medical imaging, and sparse eigensolvers. Also use when you see CPU-bound Python code (loops, large arrays, ML pipelines, graph analytics, image processing) that would benefit from GPU acceleration, even if not explicitly requested.
View skillAutonomously improve a real artifact (code, training recipe, agent harness, data pipeline, prompt) against an objective and an evaluator, using Hypothesis Tree Refinement (HTR) from the Arbor paper. Use this whenever someone wants to iteratively optimize something over many experiments without overfitting — e.g. "get my model's eval score up", "improve this agent/harness", "tune this pipeline", "beat the baseline on this benchmark", "run a search over approaches and keep the best", "do an MLE-bench / Kaggle-style optimization", or any long-horizon "make this artifact better and don't just memorize the dev set" task. Trigger it even when the user doesn't say "Arbor" or "hypothesis tree" but describes repeated experiment-and-evaluate loops, branching exploration of competing ideas, or worries about a dev/test gap. Runs Claude itself as the coordinator with subagent executors in isolated git worktrees; for the standalone `arbor` CLI tool see references/arbor-upstream.md.
View skillWrite publication-ready ML/AI papers for NeurIPS, ICML, ICLR, ACL, AAAI, COLM. Use when drafting papers from research repos, structuring arguments, verifying citations, or preparing camera-ready submissions. For systems venues (OSDI, NSDI, ASPLOS, SOSP), use systems-paper-writing instead.
View skill