Learning From Correctness Without Prompting Makes LLM Efficient Reasoner
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909261268779008 |
|---|---|
| author | Yao, Yuxuan Wu, Han Guo, Zhijiang Zhou, Biyan Gao, Jiahui Luo, Sichun Hou, Hanxu Fu, Xiaojin Song, Linqi |
| author_facet | Yao, Yuxuan Wu, Han Guo, Zhijiang Zhou, Biyan Gao, Jiahui Luo, Sichun Hou, Hanxu Fu, Xiaojin Song, Linqi |
| contents | Large language models (LLMs) have demonstrated outstanding performance across various tasks, yet they still exhibit limitations such as hallucination, unfaithful reasoning, and toxic content. One potential approach to mitigate these issues is learning from human or external feedback (e.g. tools). In this paper, we introduce an intrinsic self-correct reasoning framework for LLMs that eliminates the need for human feedback, external tools, and handcraft prompts. The proposed framework, based on a multi-step reasoning paradigm \textbf{Le}arning from \textbf{Co}rrectness (\textsc{LeCo}), improves reasoning performance without needing to learn from errors. This paradigm prioritizes learning from correct reasoning steps, and a unique method to measure confidence for each reasoning step based on generation logits. Experimental results across various multi-step reasoning tasks demonstrate the effectiveness of the framework in improving reasoning performance with reduced token consumption. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_19094 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Learning From Correctness Without Prompting Makes LLM Efficient Reasoner Yao, Yuxuan Wu, Han Guo, Zhijiang Zhou, Biyan Gao, Jiahui Luo, Sichun Hou, Hanxu Fu, Xiaojin Song, Linqi Computation and Language Large language models (LLMs) have demonstrated outstanding performance across various tasks, yet they still exhibit limitations such as hallucination, unfaithful reasoning, and toxic content. One potential approach to mitigate these issues is learning from human or external feedback (e.g. tools). In this paper, we introduce an intrinsic self-correct reasoning framework for LLMs that eliminates the need for human feedback, external tools, and handcraft prompts. The proposed framework, based on a multi-step reasoning paradigm \textbf{Le}arning from \textbf{Co}rrectness (\textsc{LeCo}), improves reasoning performance without needing to learn from errors. This paradigm prioritizes learning from correct reasoning steps, and a unique method to measure confidence for each reasoning step based on generation logits. Experimental results across various multi-step reasoning tasks demonstrate the effectiveness of the framework in improving reasoning performance with reduced token consumption. |
| title | Learning From Correctness Without Prompting Makes LLM Efficient Reasoner |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2403.19094 |