Unlocking Anticipatory Text Generation: A Constrained Approach for Large Language Models Decoding
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916422999867392 |
|---|---|
| author | Tu, Lifu Yavuz, Semih Qu, Jin Xu, Jiacheng Meng, Rui Xiong, Caiming Zhou, Yingbo |
| author_facet | Tu, Lifu Yavuz, Semih Qu, Jin Xu, Jiacheng Meng, Rui Xiong, Caiming Zhou, Yingbo |
| contents | Large Language Models (LLMs) have demonstrated a powerful ability for text generation. However, achieving optimal results with a given prompt or instruction can be challenging, especially for billion-sized models. Additionally, undesired behaviors such as toxicity or hallucinations can manifest. While much larger models (e.g., ChatGPT) may demonstrate strength in mitigating these issues, there is still no guarantee of complete prevention. In this work, we propose formalizing text generation as a future-constrained generation problem to minimize undesirable behaviors and enforce faithfulness to instructions. The estimation of future constraint satisfaction, accomplished using LLMs, guides the text generation process. Our extensive experiments demonstrate the effectiveness of the proposed approach across three distinct text generation tasks: keyword-constrained generation (Lin et al., 2020), toxicity reduction (Gehman et al., 2020), and factual correctness in question-answering (Gao et al., 2023). |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2312_06149 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Unlocking Anticipatory Text Generation: A Constrained Approach for Large Language Models Decoding Tu, Lifu Yavuz, Semih Qu, Jin Xu, Jiacheng Meng, Rui Xiong, Caiming Zhou, Yingbo Computation and Language Artificial Intelligence Large Language Models (LLMs) have demonstrated a powerful ability for text generation. However, achieving optimal results with a given prompt or instruction can be challenging, especially for billion-sized models. Additionally, undesired behaviors such as toxicity or hallucinations can manifest. While much larger models (e.g., ChatGPT) may demonstrate strength in mitigating these issues, there is still no guarantee of complete prevention. In this work, we propose formalizing text generation as a future-constrained generation problem to minimize undesirable behaviors and enforce faithfulness to instructions. The estimation of future constraint satisfaction, accomplished using LLMs, guides the text generation process. Our extensive experiments demonstrate the effectiveness of the proposed approach across three distinct text generation tasks: keyword-constrained generation (Lin et al., 2020), toxicity reduction (Gehman et al., 2020), and factual correctness in question-answering (Gao et al., 2023). |
| title | Unlocking Anticipatory Text Generation: A Constrained Approach for Large Language Models Decoding |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2312.06149 |