Research map / Large language models / Agents, code and math
SwiftSage: A Generative Agent with Fast and Slow Thinking for Complex Interactive Tasks
Bill Yuchen Lin, Yicheng Fu, Karina Yang, Faeze Brahman, Shiyu Huang, Chandra Bhagavatula, Prithviraj Ammanabrolu, Yejin Choi, Xiang Ren
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add the Sage (GPT-4 planning and grounding) module to Swift-only agent (ablation) | SwiftSage · ScienceWorld (30 tasks, overall) | average score: 49.22 -> 62.22 | Improved |
| add Sage module (GPT-3.5) to Swift-only on short tasks (ablation) | SwiftSage · ScienceWorld short tasks | average score: 78.68 -> 72.81 | Hurt |
| add Sage module (GPT-3.5) to Swift-only on medium tasks (ablation) | SwiftSage · ScienceWorld medium tasks | average score: 32.90 -> 55.34 | Improved |
| add Sage module (GPT-3.5) to Swift-only on long tasks (ablation) | SwiftSage · ScienceWorld long tasks | average score: 35.55 -> 57.99 | Improved |
| include answers to Q1-Q3 in the grounding stage (ablation) | Sage module · ScienceWorld short tasks | average score: about 2 points | Improved |
| swap GPT-4 for GPT-3.5-turbo as the LLM | SwiftSage · ScienceWorld (30 tasks, overall) | average score: 84.68 -> 62.22 | Hurt |
| swap GPT-4 for GPT-3.5-turbo as the LLM | ReAct · ScienceWorld (30 tasks, overall) | average score: 36.43 -> 19.76 | Hurt |
| fast-slow SwiftSage instead of per-step LLM prompting (ReAct) | GPT-4 agents · ScienceWorld (30 tasks, overall) | tokens per action: 1971.03 -> 757.07 | Improved |
| balanced T5-large Swift module with sliding-window history instead of TDT (11B) | behavior cloning agent · ScienceWorld (30 tasks, overall) | average score: 14.66 -> 49.22 | Improved |
Closest papers by meaning
- THOUGHT PROPAGATION: AN ANALOGICAL APPROACH TO COMPLEX REASONING WITH LARGE LANGUAGE MODELS (ICLR 2024 · similarity 0.87)
- Eliciting Reasoning in Language Models with Cognitive Tools (NeurIPS 2025 · similarity 0.86)
- Unlocking the Power of Multi-Agent LLM for Reasoning: From Lazy Agents to Deliberation (ICLR 2026 · similarity 0.86)
- MemGen: Weaving Generative Latent Memory for Self-Evolving Agents (ICLR 2026 · similarity 0.86)
- ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory (ICLR 2026 · similarity 0.86)
- Large Language Models Think Too Fast To Explore Effectively (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
