SakanaAI/AI-Scientist-v3

选用指南

AI 根据 README 整理 · 2026/10/17 05:56:19

适用场景

适用于从任务描述自动生成实验、消融和论文的自动化科学发现流程。

查看 README 依据

<a href="https://arxiv.org/pdf/2610.18933"><img src="https://img.shields.io/badge/Paper-PDF-red.svg" alt="Paper PDF"></a> <a href="LICENSE"><img src="https://img.shields.io/badge/License-Apache_2.0-blue.svg" alt="License: Apache 2.0"></a> </p> Agentic Scientist treats coding agents as AI scientists. Starting from a `task.md`, the agents propose ideas, run experiments, evaluate outcomes against the task rubric, and iterate toward stronger programs. The pipeline then runs baselines and ablations, a sealed final evaluation, post-hoc analyses, and

使用方式

通过命令行使用，主要命令为 `agsci run`，支持 `--dry-run` 预览，并可指定阶段和参数。

查看 README 依据

Point `agsci run` at a task directory. This runs the default six phases (task drafting → investigation → ablations → final evaluation → post-hoc analysis → paper) on the CPU-only circle-packing example: ```bash agsci run examples/circpack ``` Add `--dry-run` to validate the configuration and preview the exact commands without running anything. Every option is a Hydra `key=value` override; `agsci run --help` lists them all. > **Note** — Agentic Scientist drives `claude`, `codex`, `cursor`, `opencode`, > `gemini`, and `pi` through

# Continue after investigation with ablations, post-hoc analysis, and the paper: agsci run examples/circpack execution.phases=ablations,posthoc,paper \ paths.run_dir=results/circpack/claude/20260331_094917 # Rewrite only the paper: agsci run examples/circpack execution.phases=paper \ paths.run_dir=results/circpack/claude/20260331_094917 ``` ### 4. Send the paper to an external reviewer Adding the `review` phase sends the complete manuscript to the OpenAI-backed reviewer. Review failures are recorded but never discard the paper: ```bash

部署要求

需要 Python ≥ 3.11 和 uv，以及 Docker（默认运行环境）。需要 API 密钥（如 ANTHROPIC_API_KEY、OPENAI_API_KEY）供代理使用。GPU 仅特定任务需要。

查看 README 依据

> permissions (Headless `--allow yolo`). They can run arbitrary commands and > inherit the API keys you give them, and the task files and papers they read > steer what they do. Keep runs inside the Docker image rather than on your > workstation, only run tasks and sources you trust, and review run directories > before sharing them: agent logs can echo document contents and paths. ### Prerequisites - **Python ≥ 3.11 with [uv](https://docs.astral.sh/uv/)**. `uv sync` creates `.venv` and installs the `agsci` command: ```bash curl -LsSf https://astral.sh/uv/install.sh | sh uv sync

source .venv/bin/activate ``` - **API keys** in a `.env` file at the repo root (`ANTHROPIC_API_KEY`, `OPENAI_API_KEY`, `GEMINI_API_KEY`, ... whichever your agent needs). The pipeline mounts it into the agent containers. Never commit it. - **Agent CLIs** are already inside the image. For native (non-containerized) runs, install the CLI of the agent you use plus Git and Node.js; versions and commands are in [docs/install.md](docs/install.md). - **GPUs** are required only by the GPU task presets. The pipeline discovers

已知限制

从研究简报起草任务仍处于实验阶段，对大型任务可能无法生成符合完整流水线契约的脚手架。

查看 README 依据

A task directory holding only a `task.md` brief makes the drafting phase write the initial program, evaluators, and task contract itself: ```bash agsci run examples/circpack_task_drafting execution.phases=task_drafting ``` This works for simple tasks such as circle packing but is not yet robust: for larger tasks like the world-model or tabular distribution-shift examples it may fail to produce a scaffold that matches the full pipeline contract. Start from a shipped scaffold for anything beyond a small task, and check the drafting audit under `task_drafting/` when a draft is rejected. See