Behavior completely inconsistent with the 7b model?
#5
by
EntropyYue
- opened
When I use the 7b distill version, its performance is similar to the version on the official website, but the 14b distill version's thinking is shorter and not very complex
EntropyYue
changed discussion status to
closed
Alright, it still looks pretty much the same; it might be because I haven't tested it enough