Improving Open-Ended Text Generation via Adaptive Decoding
This work addresses text generation quality for users of language models, presenting an incremental improvement over existing decoding methods.
The paper tackles the problem of open-ended text generation by introducing adaptive decoding, which dynamically selects candidate tokens based on an entropy-based confidence metric, resulting in improved balance between diversity and coherence and human-preferred text generation.
Current language models decode text token by token according to probabilistic distribution, and determining the appropriate candidates for the next token is crucial to ensure generation quality. This study introduces adaptive decoding, a mechanism that dynamically empowers language models to ascertain a sensible candidate set during generation. Specifically, we introduce an entropy-based metric called confidence and conceptualize determining the optimal candidate set as a confidence-increasing process. The rationality of including a token in the candidate set is assessed by leveraging the increment of confidence. Experimental results reveal that our method balances diversity and coherence well. The human evaluation shows that our method can generate human-preferred text. Additionally, our method can potentially improve the reasoning ability of language models.