Are there any (small, ~10B) models that you would say are a good collaborator?
Most of the new ~30B (and now ~10B, thankfully for my GPU) models we see score really high on benchmarks, but I feel like they don't push back on dumb ideas enough. I think most people don't being like told by an LLM that the premise is flawed but I certainly do. In my opinion they are optimised for like one-shotting stuff, but I don't want it to do that. Especially from like a 10B model.
评论
?
参与讨论