This actually isn’t limited to code, now I am seeing it happen on any long-ish response from the LLM.
At some point (even just generating raw markdown) GPT-4o starts outputting bad linebreaks.
This actually isn’t limited to code, now I am seeing it happen on any long-ish response from the LLM.
At some point (even just generating raw markdown) GPT-4o starts outputting bad linebreaks.