Hacker News new | past | comments | ask | show | jobs | submit login

Not necessarily true anymore. As a counterexample, consider CodeLlama 33b, which is quite good (and which has replaced GPT-4 for my coding assistant needs).

OpenAI's models are likely to remain the best, but I see open models catching up and becoming "good enough." Why pay for GPT-4 when I can run a model locally for free? (Barring the initial capital cost of a GPU; and not even this if you're, say, using a Macbook)




CodeLlama spit out some of the wildest garbage when I tested it, even against GPT-3.5. Which tasks have you found it to perform well?


What's CodeLlama 33b? I couldn't find anything regarding it online



Off by 1B error


Just a 1B mistake


^-^


Whoops, sorry. 34b :)


Not 100% sure but I believe Llama [1] is a LLM created by meta. Code Llama is probably one tailored as a coding assistant

[1] https://ai.meta.com/llama/




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: