Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2509.20820 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916969801842688 |
|---|---|
| author | Honda, Ukyo Murakami, Soichiro Zhang, Peinan |
| author_facet | Honda, Ukyo Murakami, Soichiro Zhang, Peinan |
| contents | Recent advances in large language models (LLMs) enable effective in-context learning (ICL) with many-shot examples, but at the cost of high computational demand due to longer input tokens. To address this, we propose cheat-sheet ICL, which distills the information from many-shot ICL into a concise textual summary (cheat sheet) used as the context at inference time. Experiments on challenging reasoning tasks show that cheat-sheet ICL achieves comparable or better performance than many-shot ICL with far fewer tokens, and matches retrieval-based ICL without requiring test-time retrieval. These findings demonstrate that cheat-sheet ICL is a practical alternative for leveraging LLMs in downstream tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_20820 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Distilling Many-Shot In-Context Learning into a Cheat Sheet Honda, Ukyo Murakami, Soichiro Zhang, Peinan Computation and Language Recent advances in large language models (LLMs) enable effective in-context learning (ICL) with many-shot examples, but at the cost of high computational demand due to longer input tokens. To address this, we propose cheat-sheet ICL, which distills the information from many-shot ICL into a concise textual summary (cheat sheet) used as the context at inference time. Experiments on challenging reasoning tasks show that cheat-sheet ICL achieves comparable or better performance than many-shot ICL with far fewer tokens, and matches retrieval-based ICL without requiring test-time retrieval. These findings demonstrate that cheat-sheet ICL is a practical alternative for leveraging LLMs in downstream tasks. |
| title | Distilling Many-Shot In-Context Learning into a Cheat Sheet |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2509.20820 |