Hacker News new | ask | show | jobs
by IceWreck 1150 days ago
Just tested it. I tried to get it to write a conclusion after giving it my report and while is was correct, it was kind of silly, a far cry from chatgpt. But again, this is the 7B variant and its open source.
1 comments

This is the Alpha of the 7B variant only partially through its initial training set (800B tokens out of 1.5 Trillion)