Request for small distill models that can run on laptop
Request for small distill models that can run on laptop
That's just a request to finetune smaller models on V3's dataset.
What made "distills" of R1 so great is that those "distilled" models were never tuned on "thinking" data. V3 however isn't a thinking model, so it's unlikely that its dataset would improve the already existing models in any way, unless you like Deepseek V3's writing style, that is.
That's just a request to finetune smaller models on V3's dataset.
What made "distills" of R1 so great is that those "distilled" models were never tuned on "thinking" data. V3 however isn't a thinking model, so it's unlikely that its dataset would improve the already existing models in any way, unless you like Deepseek V3's writing style, that is.
Scratch that "distill" word from that request then and we have "Request for small models that can run on laptop". Now that's something I can always sign up.