News for hackers. Stuff that matters.

A comment on Chain-of-Thought Reasoning Without Prompting. Parent comment

Re:Chain-of-Thought Reasoning Without Prompting by reqo on Saturday February 17, 2024 @03:48PM (#39413415)
Isn’t that what the softmax layer is doing? The token with highest probability among all the available tokens in the model dictionary is chosen as the next token!
[ Parent | on HN ]
Re:Chain-of-Thought Reasoning Without Prompting by danielmarkbruce on Saturday February 17, 2024 @07:33PM (#39414902)
no. Softmax layer produces a distribution. What you do with that is up to you. There are numerous ways to choose from that distribution.
[ Parent | on HN ]