| ReST meets ReAct: Self-Improvement for Multi-Step Reasoning LLM Agent | | 76.1 | ReST meets ReAct (PaLM 2-L + Google Search) | 2023-12-15 |
| Answering Questions by Meta-Reasoning over Multiple Chains of Thought | ✓ Link | 66.5 | MCR (code-davinci-002) + Google Search | 2023-04-25 |
| Making Retrieval-Augmented Language Models Robust to Irrelevant Context | ✓ Link | 62.7 | RALM (LLaMA2-13B + Google Search) | 2023-10-02 |
| Measuring and Narrowing the Compositionality Gap in Language Models | ✓ Link | 60.0 | Self-ask (GPT-3; davinci-002) + Google Search | 2022-10-07 |
| Measuring and Narrowing the Compositionality Gap in Language Models | ✓ Link | 57.6 | Self-ask (GPT-3; davinci-002) | 2022-10-07 |
| Measuring and Narrowing the Compositionality Gap in Language Models | ✓ Link | 46.4 | Chain-of-Thought (GPT-3; davinci-002) | 2022-10-07 |
| FireAct: Toward Language Agent Fine-tuning | | 44.0 | FireAct | 2023-10-09 |
| Measuring and Narrowing the Compositionality Gap in Language Models | ✓ Link | 17.6 | Direct Prompting (GPT-3; davinci-002) | 2022-10-07 |
| Measuring and Narrowing the Compositionality Gap in Language Models | ✓ Link | 0 | Google Search | 2022-10-07 |