Research map / Large language models / Agents, code and math
SkillGen: Learning Domain Skills for In-Context Sequential Decision Making
Ruomeng Ding, Wei Cheng, Minglai Shao, Chen Zhao
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 |
|---|---|---|---|
| use SkillGen skill prompts instead of Synapse 3-shot expert trajectories | Qwen2.5-7B-Instruct · ALFWorld | Success Rate (%): 19.4 -> 55.2 | Improved |
| use SkillGen instead of Synapse 3-shot | GPT-4o-mini · ALFWorld | Success Rate (%): 17.8 -> 29.8 | Improved |
| use SkillGen instead of baselines | Qwen2.5-7B-Instruct · BabyAI | Grounding Rate (%): 78.6 vs 66.7 | Mixed |
| use focused golden segment instead of a fixed or retrieved full demonstration (ablation) | SkillGen · ALFWorld (Qwen-Turbo) | Success Rate: +17.2% over Fixed, +3.7% over Retrieved | Improved |
| add step-wise skills to the golden segment (ablation) | SkillGen · ALFWorld | Success Rate: +8.1% | Improved |
| add step-wise retrieval instead of injecting all domain skills (ablation) | SkillGen · ScienceWorld | Success Rate: +5.7% | Improved |
| inject all domain skills without retrieval (ablation) | SkillGen · ScienceWorld | Progress Rate: +3.6% PR | Improved |
| add TD-based credit assignment over segment retrieval with sparse progress signal (ablation) | SkillGen (Qwen2.5-7B-Instruct) · ALFWorld | Success Rate: +12.8 SR | Improved |
| add TD-based credit assignment over segment retrieval (ablation) | SkillGen (GPT-4o-mini) · ScienceWorld | Success Rate: 16.8% increase | Improved |
| use subgoal annotations as weak supervision rather than only fully successful trajectories (ablation) | SkillGen (GPT-4o-mini) · ScienceWorld | Success Rate (%): 32.4 -> 40.2 | Improved |
| increase sampled trajectories per task from 6 to 9 or 12 (ablation) | SkillGen · ALFWorld, ScienceWorld | Success Rate | No clear effect |
| retrieve more distinct actions (s=3, k=1) instead of s=1 (ablation) | SkillGen (Qwen2.5-7B-Instruct) · ScienceWorld | Success Rate (%): 14.4% SR | Hurt |
Closest papers by meaning
- Scaling Prompt Synthesis for Large Language Model Reasoning (ICML 2026 · similarity 0.88)
- StrategyLLM: Large Language Models as Strategy Generators, Executors, Optimizers, and Evaluators for Problem Solving (NeurIPS 2024 · similarity 0.88)
- How to Auto-optimize Prompts for Domain Tasks? Adaptive Prompting and Reasoning through Evolutionary Domain Knowledge Adaptation (NeurIPS 2025 · similarity 0.87)
- Reward Is Enough: LLMs Are In-Context Reinforcement Learners (ICLR 2026 · similarity 0.87)
- Self-Generated In-Context Examples Improve LLM Agents for Sequential Decision-Making Tasks (NeurIPS 2025 · similarity 0.87)
- PromptAgent: Strategic Planning with Language Models Enables Expert-level Prompt Optimization (ICLR 2024 · similarity 0.87)
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.
