Learning Task Decomposition to Assist Humans in Competitive Programming
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915177876684800 |
|---|---|
| author | Wen, Jiaxin Zhong, Ruiqi Ke, Pei Shao, Zhihong Wang, Hongning Huang, Minlie |
| author_facet | Wen, Jiaxin Zhong, Ruiqi Ke, Pei Shao, Zhihong Wang, Hongning Huang, Minlie |
| contents | When using language models (LMs) to solve complex problems, humans might struggle to understand the LM-generated solutions and repair the flawed ones. To assist humans in repairing them, we propose to automatically decompose complex solutions into multiple simpler pieces that correspond to specific subtasks. We introduce a novel objective for learning task decomposition, termed assistive value (AssistV), which measures the feasibility and speed for humans to repair the decomposed solution. We collect a dataset of human repair experiences on different decomposed solutions. Utilizing the collected data as in-context examples, we then learn to critique, refine, and rank decomposed solutions to improve AssistV. We validate our method under competitive programming problems: under 177 hours of human study, our method enables non-experts to solve 33.3\% more problems, speeds them up by 3.3x, and empowers them to match unassisted experts. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_04604 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Learning Task Decomposition to Assist Humans in Competitive Programming Wen, Jiaxin Zhong, Ruiqi Ke, Pei Shao, Zhihong Wang, Hongning Huang, Minlie Computation and Language Programming Languages When using language models (LMs) to solve complex problems, humans might struggle to understand the LM-generated solutions and repair the flawed ones. To assist humans in repairing them, we propose to automatically decompose complex solutions into multiple simpler pieces that correspond to specific subtasks. We introduce a novel objective for learning task decomposition, termed assistive value (AssistV), which measures the feasibility and speed for humans to repair the decomposed solution. We collect a dataset of human repair experiences on different decomposed solutions. Utilizing the collected data as in-context examples, we then learn to critique, refine, and rank decomposed solutions to improve AssistV. We validate our method under competitive programming problems: under 177 hours of human study, our method enables non-experts to solve 33.3\% more problems, speeds them up by 3.3x, and empowers them to match unassisted experts. |
| title | Learning Task Decomposition to Assist Humans in Competitive Programming |
| topic | Computation and Language Programming Languages |
| url | https://arxiv.org/abs/2406.04604 |