Research map / Large language models / AI and society
OpenAGI: When LLM Meets Domain Experts
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 RLTF on top of fine-tuning (ablation) | LLaMA-2-13B · OpenAGI benchmark | Overall score: 0.2967 -> 0.3735 | Helped |
| add RLTF on top of fine-tuning (ablation) | Vicuna-7B · OpenAGI benchmark | Overall score: 0.2239 -> 0.3018 | Helped |
| add RLTF on top of fine-tuning (ablation) | Flan-T5-Large · OpenAGI benchmark | Overall score: 0.2105 -> 0.2544 | Helped |
| fine-tune instead of few-shot prompting (ablation) | LLaMA-2-13B · OpenAGI benchmark | Overall score: 0.1817 -> 0.2967 | Helped |
| fine-tune instead of few-shot prompting (ablation) | Vicuna-7B · OpenAGI benchmark | Overall score: 0.1428 -> 0.2239 | Helped |
| add few-shot demonstrations vs zero-shot (ablation) | GPT-4 · OpenAGI benchmark | Overall score: 0.2378 -> 0.5281 | Helped |
| add few-shot demonstrations vs zero-shot (ablation) | GPT-3.5-turbo · OpenAGI benchmark | Overall score: 0.1450 -> 0.3772 | Helped |
| add few-shot demonstrations vs zero-shot (ablation) | Flan-T5-Large · OpenAGI benchmark | Overall score: 0.0 -> 0.0829 | Helped |
| add detailed model descriptions to zero-shot prompt (Prompt-2) (ablation) | GPT-4 · OpenAGI benchmark | Overall score: 0.2378 -> 0.3717 | Helped |
| add detailed model descriptions to zero-shot prompt (Prompt-2) (ablation) | Claude-2 · OpenAGI benchmark | Overall score: 0.2064 -> 0.3430 | Helped |
| add detailed model descriptions to zero-shot prompt (Prompt-2) (ablation) | LLaMA-2-13B · OpenAGI benchmark | Overall score: 0.1533 -> 0.0572 | Hurt |
| add detailed model descriptions to zero-shot prompt (Prompt-2) (ablation) | Vicuna-7B · OpenAGI benchmark | Overall score: 0.0739 -> 0.0884 | Mixed |
Closest papers by meaning
- Mixing Expert Knowledge: Bring Human Thoughts Back To the Game of Go (NeurIPS 2025 · similarity 0.88)
- Product of Experts with LLMs: Boosting Performance on ARC Is a Matter of Perspective (ICML 2025 · similarity 0.87)
- SuperGPQA: Scaling LLM Evaluation across 285 Graduate Disciplines (NeurIPS 2025 · similarity 0.87)
- Large Language Models Think Too Fast To Explore Effectively (NeurIPS 2025 · similarity 0.87)
- Advancing LLM Reasoning with Natural Language and Numerical Feedback (ICML 2026 · similarity 0.86)
- AGI-Elo: How Far Are We From Mastering A Task? (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.