3 ms·
> Keep in mind that all LLMs are trained first on text and then fine-tuned on code. No, they are trained on a mixture of text and code from the start.
by in-silico 5mo ago
> Keep in mind that all LLMs are trained first on text and then fine-tuned on code.
No, they are trained on a mixture of text and code from the start.