model choice is visible but endpoint unreachable

[theoden8]

Hi there,

Thank you for a wonderful piece of software, it has a very friendly interface and a great take on usability.

I am using ollama with ollama_proxy_server and I see / requests every 5 seconds (it is spamming in the logs) but the completion request is not visible. This proxy works fine with some other GUI software for LLMs. Model selection is detected.

On MacOS:

On iOS:

ollama_proxy_server logs:

127.0.0.1 - -  "GET /api/tags HTTP/1.0" - -
127.0.0.1 - -  "GET /api/tags HTTP/1.0" 200 -
127.0.0.1 - -  code 501, message Unsupported method ('HEAD')
127.0.0.1 - -  "HEAD / HTTP/1.0" 501 -
127.0.0.1 - -  code 501, message Unsupported method ('HEAD')
127.0.0.1 - -  "HEAD / HTTP/1.0" 501 -
127.0.0.1 - -  code 501, message Unsupported method ('HEAD')
127.0.0.1 - -  "HEAD / HTTP/1.0" 501 -

[gluonfield]

Thanks for sharing, I will investigate!

[theoden8]

/api/tags is not authenticated, it seems, if that's any help.