preprint · arXiv (Cornell University) · 2024

Impact of Decoding Methods on Human Alignment of Conversational LLMs

Shaz Furniturewala, Kokil Jaidka, Yashvardhan Sharma

Cited by 0Open access

Abstract

To be included into chatbot systems, Large language models (LLMs) must be aligned with human conversational conventions. However, being trained mainly on web-scraped data gives existing LLMs a voice closer to informational text than actual human speech. In this paper, we examine the effect of decoding methods on the alignment between LLM-generated and human conversations, including Beam Search, Top K Sampling, and Nucleus Sampling. We present new measures of alignment in substance, style, and psychometric orientation, and experiment with two conversation datasets. Our results provide subtle insights: better alignment is attributed to fewer beams in Beam Search and lower values of P in Nucleus Sampling. We also find that task-oriented and open-ended datasets perform differently in terms of alignment, indicating the significance of taking into account the context of the interaction.

Cite this paper

APA 7

Furniturewala, S., Jaidka, K., & Sharma, Y. (2024). Impact of Decoding Methods on Human Alignment of Conversational LLMs. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2407.19526

MLA 9

Furniturewala, Shaz, et al. “Impact of Decoding Methods on Human Alignment of Conversational LLMs.” arXiv (Cornell University), 2024, https://doi.org/10.48550/arxiv.2407.19526.

Chicago

Furniturewala, Shaz, Kokil Jaidka, and Yashvardhan Sharma. 2024. “Impact of Decoding Methods on Human Alignment of Conversational LLMs.” arXiv (Cornell University). https://doi.org/10.48550/arxiv.2407.19526.

Harvard

Furniturewala, S., Jaidka, K. and Sharma, Y. (2024) ‘Impact of Decoding Methods on Human Alignment of Conversational LLMs’, arXiv (Cornell University). Available at: https://doi.org/10.48550/arxiv.2407.19526.

BibTeX

@misc{furniturewala2024impact,
  title = {{Impact of Decoding Methods on Human Alignment of Conversational LLMs}},
  author = {Furniturewala, Shaz and Jaidka, Kokil and Sharma, Yashvardhan},
  journal = {arXiv (Cornell University)},
  year = {2024},
  doi = {10.48550/arxiv.2407.19526}
}

Made from the index record. Check names, capitalisation and page numbers against the paper. Need another style? AI for Research has over 10,000.