Ask HN: Another front end for open LLM?
One of the biggest pain point I had with existing LLM front end are that they require some technical skills and a powerful pc (installation, downloading model, and make sure all the dependencies works as expected). Which gives me an idea. Why not create an online community like llm front end where you can share your character, use already deployed model, and make it accessible any where (like responsive, accessible everywhere with internet access)
Think Civitai + hugging face inference endpoint + kobold cpp. Slap a responsive and fast UI. (Many model on huge g face cannot be use as is and need some sort of deployment)
And for inference and compute you can use your own API key from Replicate AI, together AI, or any other poplar inference platform. You pay only what you use to the cloud provider (many also have free credit) and it will deploy the model you want to use for you and your data will all keep to yourself.
That way we can make LLM and SD image model more accessible. Idk does something like this resonate with people?