- cross-posted to:
- ai_@lemmy.world
You must log in or # to comment.
And yet, I still won’t
They’re using fine-tuning on top of existing models. Still useful, but the headline implies creating foundation models, which is still the domain of big iron.
“Can”
The fuck does 70B mean? 70 bytes?
70 billion parameters. It’s standard notation in the context of LLMs.


