Replying to Avatar Neigsndoig IQ 120

What LLM do you guys use as the base? I hope it's not one that's using Common Crawl, since that's disgusting slop (and plagiarized to boot).
Avatar
Marks 8mo ago 💬 1

Llama 3.3 70B. Currently exploring other models to give users options.

Reply to this note

Please Login to reply.

Discussion

Avatar
Neigsndoig IQ 120 8mo ago 💬 1

That sounds like a Common Crawl model, which is not a good dataset to use AI on. At least it's "open-source" (It's ethical source, let's be real here).

Maybe I could interest you in taking a look at non-Common Crawl models like DeepSeek R1 if reasoning modeling is being planned?

Avatar
Marks 8mo ago

R1 is one of the models we are exploring (R2 might be out soon too)