# Getting a whole distribution for GPT next token

**URL:** </static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fgetting-a-whole-distribution-for-gpt-next-token%2F10568%3E%5C
**Category:** 🤗Transformers\
**Created:** [October 6, 2021, 3:14pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fgetting-a-whole-distribution-for-gpt-next-token%2F10568 "2021-10-06T15:14:27Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Keverino](https://avatars.discourse-cdn.com/v4/letter/k/b2d939/32.png) [@Keverino](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Fu%2FKeverino)%5C
**Post date:** [October 6, 2021, 3:14pm UTC](/static-proxy?url=https%3A%2F%2Fdiscuss.huggingface.co%2Ft%2Fgetting-a-whole-distribution-for-gpt-next-token%2F10568%2F1 "2021-10-06T15:14:27Z")

</div>

Hello,

I am working with GPT-J.  
Generating text for some input via model.generate() is easy and straightforward.

What if I want to have the top 5 or 10 predictions for each word with its probability scores?  
Is there a simple way to do this? I feel like the sampling methods in generate() don’t allow this…

Thanks in advance!