Incompatibility with llama.cpp API: /api/generate Endpoint Not Supported

[vanabel]

Details:
I am using llama.cpp with GPU support for my projects. However, I found that the /api/generate endpoint, which might be expected by Copilot for Obsidian, is not supported by llama.cpp. Instead, the correct endpoint for generating text is /v1/completions.

This issue has already been reported to Ollama-Logseq. For more details, please refer to their report. You might want to update the plugin to accommodate the correct API path in llama.cpp.

Reference:

• For the correct API documentation, see: llama.cpp API Documentation.

Thank you for looking into this.

[logancyang]

I suppose you are using Copilot for ob with Ollama? Without a screenshot of your query and dev console I don't really follow what the issue is.

Ok you are directly using it with Llama.cpp. Copilot doesn't integrate with llama.cpp directly, it has option for either Ollama or LM Studio for local models.

This issue sounds like a bug report but is actually a feature request to support llama.cpp?