Hacker News new | ask | show | jobs
by krainboltgreene 985 days ago
Why? Are not grammatical errors and idiosyncrasies part of the data set that chatGPT is trained on and thus also a predictable set of tokens?
1 comments
btilly 985 days ago
Such errors are rare enough that they are almost never the most likely next token. Therefore ChatGPT only makes such mistakes when it is given good reason to do so.
link
krainboltgreene 984 days ago
ChatGPT cannot reason.
link