9 ms·
I've been diligently trying to use Gemini 1.5 Pro, and it is not even on the level of Llama3-70B. I really hope Gemini improves, even if it gets reduced context
by stan_kirdey 2y ago
I've been diligently trying to use Gemini 1.5 Pro, and it is not even on the level of Llama3-70B. I really hope Gemini improves, even if it gets reduced context length.
- ukuina 2y agoFAIR really swung for the fences with Llama3. It's a very impressive model, but the 8K context size is quite limiting for most use-cases.