Following Length Constraints in Instructions
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_ | 1866910501046321152 |
|---|---|
| author | Yuan, Weizhe Kulikov, Ilia Yu, Ping Cho, Kyunghyun Sukhbaatar, Sainbayar Weston, Jason Xu, Jing |
| author_facet | Yuan, Weizhe Kulikov, Ilia Yu, Ping Cho, Kyunghyun Sukhbaatar, Sainbayar Weston, Jason Xu, Jing |
| contents | Aligned instruction following models can better fulfill user requests than their unaligned counterparts. However, it has been shown that there is a length bias in evaluation of such models, and that training algorithms tend to exploit this bias by learning longer responses. In this work we show how to train models that can be controlled at inference time with instructions containing desired length constraints. Such models are superior in length instructed evaluations, outperforming standard instruction following models such as GPT4, Llama 3 and Mixtral. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_17744 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Following Length Constraints in Instructions Yuan, Weizhe Kulikov, Ilia Yu, Ping Cho, Kyunghyun Sukhbaatar, Sainbayar Weston, Jason Xu, Jing Computation and Language Aligned instruction following models can better fulfill user requests than their unaligned counterparts. However, it has been shown that there is a length bias in evaluation of such models, and that training algorithms tend to exploit this bias by learning longer responses. In this work we show how to train models that can be controlled at inference time with instructions containing desired length constraints. Such models are superior in length instructed evaluations, outperforming standard instruction following models such as GPT4, Llama 3 and Mixtral. |
| title | Following Length Constraints in Instructions |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2406.17744 |