There’s the ALMA models based on LLaMA 2:

https://huggingface.co/haoranxu/ALMA-13B

I’ve tried this for translating Japanese, seems pretty good: https://huggingface.co/mmnga/webbigdata-ALMA-7B-Ja-V2-gguf

No, several sources include Microsoft have said GPT 3.5 Turbo is 20B. GPT 3 was 175B, and GPT 3.5 Turbo was about 10x cheaper on the API than GPT 3 when it came out so it makes sense.

I mostly use a UI I made myself:

https://github.com/shinomakoi/AI-Messenger

Works with llama.cpp and Exllama V2, supports LLaVA, character cards and moar.

I converted and quantized this to work in llama.cpp

https://huggingface.co/nakodanei/ShareGPT4V-7B_GGUF