So I don't think Phi-2 is actually available to the public - as far as I've seen, it's just a Microsoft research paper. But I'm going to try and play with Mixtral-7B ASAP, OpenRouter is somehow offering it as a hosted API for free: https://openrouter.ai/models/mistralai/mixtral-8x7b-instruct?tab=status
Between Phi-2 and Mixtral-7B, it seems like we're entering an era of (relatively) small language models.
Yeah, definitely. Have you explored Phi-2 or Mixtral-7B? I didn't get time yet, but on my todos.
So I don't think Phi-2 is actually available to the public - as far as I've seen, it's just a Microsoft research paper. But I'm going to try and play with Mixtral-7B ASAP, OpenRouter is somehow offering it as a hosted API for free: https://openrouter.ai/models/mistralai/mixtral-8x7b-instruct?tab=status
Phi-2 is now on Hugging Face: https://huggingface.co/microsoft/phi-2
I didn't know about OpenRouter offering free API access for Mixtral-7B. Awesome! Thanks for sharing.