Closed jivanph closed 7 months ago
Hi,
We checked Section 4.1 (Experimental Setup) in the REST paper, which claims that their results are identical to those generated by autoregressive generation. Moreover, the authors of REST have provided a detailed code implementation in response to a similar query, which you can find here.
We hope this addresses your concern. Please feel free to reach out if you have any more questions or require further clarification.
Thank you so much for your response. It clarify my understanding of their methodology.
I noticed that in Table 3 you present a summary of different decoding methodologies.
In particular, you present REST as a methodology amenable to Nucleus sampling, but on reading the REST methodology paper, it's not clear how they establish that their methodology maintains the same output distribution from a LLM.