Difference between this model and the one Meta released
Hello. I know Meta-Llama-3.1-8B-Instruct-bnb-4bit is a tuned version with less param and faster inference. How about this one? It has same number of param as the original release from Meta. Is it a modified version to work with the unsloth library and has same performance as the original release?
Hello. I know Meta-Llama-3.1-8B-Instruct-bnb-4bit is a tuned version with less param and faster inference. How about this one? It has same number of param as the original release from Meta. Is it a modified version to work with the unsloth library and has same performance as the original release?
Nope - it's the exact same as the Meta LLama 3.1. We just reuploaded it! :)
I love being able to simply replace with unsloth in the url and download these without having to sign in / setup my hf key on every system I'm using.
Most of the time now I don't even have to search, I just go straight to the URL and assume you guys have re-uploaded
I love being able to simply replace with unsloth in the url and download these without having to sign in / setup my hf key on every system I'm using.
Most of the time now I don't even have to search, I just go straight to the URL and assume you guys have re-uploaded
That was not our intention - it was mostly so user's could use our fine-tuning library easier but glad you're using it in some sort other useful way! :D