Ollama not working

[diyoyo]

Hi,
I can't get the Ollama integration to work with the basic settings (http://localhost:11434 on a Mac with default Ollama app).

• The cli works perfectly fine.
• I have tried the two models described in the docs (although I just read in the Issues that for embeddings, it's not going to work).
• I get a 405 Error when using the chat from Trilium
  • If I look on Internet, it happens on certain models, and it was a bug that was supposed to be solved a long time ago (like in llama 0.3.5 or something)
  • I also read that the models needed to have the "tools" specification, so I tried some as well.
• I checked 'Expose Ollama to the network', without luck.
• If I try to fetch() from a JS Frontend script, it won't work because of CORS issues. (Note: I just converted the curl command from the main page of ollama's github, not sure it is the right thing to do.
  • I tried adding all the CORS-enabling environment varibales in a trilium launcher, but it was for the sake of having tried it, since it is supposed to help the other way around, right?
• I also tried to minify ollama.js (which was not supposed to be feasible due to the structure of the project, so ChatGPT made me cheat but I don't understand what was done) and integrate it like a class (like the renderChart.js exemple in the demo of Trilium), but here, well, it was just a matter of trying my luck. I had the same problem as I have evrytime I don't understand the structure of the inner js
  you're not allowed to use imports outside of a module
  • the idea behind it was to benefit from a streamed communication that may not be obvious using basic fetch()calls.

Anyways, I realize that I may not have tried 'JS Backend', but in the meantime, is there a known special trick?