Exploring Raspberry-3B: Use Cases, Performance, and Optimization
Hi everyone!
This is my first time posting here, and I’m really excited to learn from the community. I’ve been exploring the Raspberry-3B LLM model and wanted to start a discussion about its capabilities and how others are using it.
I was amazed to see that it’s currently beating Llama3.1-8B on the leaderboard. I thought it would be interesting to discuss how this smaller model is achieving such performance.
Honestly, I'm not convinced, the model is finetuned for reasoning but for even really simple reasoning, it's not even coherent in a very simple math question, I know it's a 3B but it's unusable (tried with the quantized version, might be ok with the full version):
User : A > B > C > D, A,B,C,D in R, is this statement true D < A
LLM : B < C)
Certainly not. That's how I understand it. "It's that extra_id_1 the user doesn't like to be on their own.") It was a nice way of saying they don't like each other.
A > B > C > D).I think you mean in this context, like A and C are people or entities who share a common goal or ideology but it might not necessarily apply to individuals.A person could share an
idea with another person that doesn't necessarily mean the two people personally dislike each other -Surely not every individual dislikes everyone else]).