Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2506.00483 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912405967077376 |
|---|---|
| author | Jan, Aviv Tahory, Dean Talmi, Omer Mokh, Omar Abo |
| author_facet | Jan, Aviv Tahory, Dean Talmi, Omer Mokh, Omar Abo |
| contents | Multi-hop questions still stump large language models (LLMs), which struggle to link information across multiple reasoning steps. We introduce Auto-Patch, a novel method that dynamically patches hidden states during inference to enhance multi-hop reasoning in LLMs. Building on the PatchScopes framework, Auto-Patch selectively modifies internal representations using a learned classifier. Evaluated on the MuSiQue dataset, Auto-Patch improves the solve rate from 18.45\% (baseline) to 23.63~$\pm$~0.7\% (3 runs), narrowing the gap to Chain-of-Thought prompting (27.44\%). Our results highlight the potential of dynamic hidden state interventions for advancing complex reasoning in LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2506_00483 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Auto-Patching: Enhancing Multi-Hop Reasoning in Language Models Jan, Aviv Tahory, Dean Talmi, Omer Mokh, Omar Abo Computation and Language Machine Learning Multi-hop questions still stump large language models (LLMs), which struggle to link information across multiple reasoning steps. We introduce Auto-Patch, a novel method that dynamically patches hidden states during inference to enhance multi-hop reasoning in LLMs. Building on the PatchScopes framework, Auto-Patch selectively modifies internal representations using a learned classifier. Evaluated on the MuSiQue dataset, Auto-Patch improves the solve rate from 18.45\% (baseline) to 23.63~$\pm$~0.7\% (3 runs), narrowing the gap to Chain-of-Thought prompting (27.44\%). Our results highlight the potential of dynamic hidden state interventions for advancing complex reasoning in LLMs. |
| title | Auto-Patching: Enhancing Multi-Hop Reasoning in Language Models |
| topic | Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2506.00483 |