OpenRaiser/NanoResearch

Academic Plotting generates publication-quality figures for ML papers from research context or experimental data. It creates architecture diagrams via Gemini and data-driven charts via matplotlib/seaborn, auto-selecting chart types and highlighting key results.

Autoresearch orchestrates end-to-end autonomous AI research projects using a two-loop architecture for rapid experiments and periodic synthesis, routing to domain-specific skills and producing papers; use when starting a research project or managing multi-hypothesis experiments.

Creative Thinking for Research applies eight cognitive science frameworks to generate novel research directions in computer science and AI. Use it when seeking genuinely novel directions via combinatorial creativity, analogical reasoning, and constraint manipulation.

Hugging Face Accelerate simplifies distributed training for PyTorch models by adding only four lines of code. Use it to run the same script on single or multiple GPUs, with mixed precision, DeepSpeed, or FSDP.

LLM Evaluation Harness evaluates LLMs across 60+ academic benchmarks using standardized prompts and metrics. Use when benchmarking model quality, comparing models, or tracking training progress.

ML Paper Writing drafts publication-ready ML/AI/systems papers for NeurIPS, ICML, ICLR, ACL, AAAI, COLM, OSDI, NSDI, ASPLOS, SOSP. Use it to turn research repos into structured papers, find and verify citations, and apply conference LaTeX templates.

ML Training Recipes provides battle-tested PyTorch training patterns for LLMs, vision, diffusion, medical imaging, and protein/drug discovery. Use it when training or fine-tuning neural networks, debugging loss spikes/OOM, choosing architectures, or optimizing GPU throughput.

NanoResearch Experiment generates a runnable Python code skeleton from an experiment blueprint, including data loaders, model stubs, training loops, evaluation harness, and ablation configs.

NanoResearch Ideation searches arXiv and Semantic Scholar for literature on a research topic, performs gap analysis, and generates novel hypotheses with justification. Use it to move from a broad topic to ranked papers and a selected promising hypothesis.

NanoResearch Planning produces an experiment blueprint from a selected research hypothesis, specifying datasets, baselines, evaluation metrics, and ablation groups.

NanoResearch Writing drafts a complete LaTeX research paper from upstream ideation, planning, and experiment outputs. Use it when you need to turn experiment results and a blueprint into a publication-ready PDF with figures, tables, and bibliography.

PEFT Fine-Tuning fine-tunes large language models by training less than 1% of parameters using LoRA, QLoRA, and 25+ adapter methods. Use it when fine-tuning 7B–70B models on consumer GPUs, when you need small task-specific adapters, or for multi-adapter serving.

Ray Data processes large datasets in parallel across CPU/GPU clusters for ML workloads. Use it for batch inference, data preprocessing, multi-modal data loading, or distributed ETL pipelines.

Research Idea Brainstorming guides researchers through structured ideation frameworks to discover high-impact research directions. Use it when exploring new problem spaces, pivoting between projects, or seeking novel angles on existing work.

SkyPilot Multi-Cloud Orchestration manages ML training and batch jobs across multiple clouds with automatic cost optimization, spot instance recovery, and distributed multi-node support. Use it when you need a unified interface for AWS, GCP, Azure, Kubernetes, and 20+ providers without vendor lock-in.

Unsloth provides expert guidance for fast fine-tuning with Unsloth, covering 2-5x faster training, 50-80% less memory, and LoRA/QLoRA optimization. Use it when working with Unsloth features, implementing solutions, or debugging training code.