suggestions

[planetadelexilio-boop]

Thanks for OpenLumara! I suggest some improvements: 1) The ability to choose the interface language. 2) An HTML previewer in the chat. 3) Finally, I usually start "sessions" with different models each time (Gemma, Qwen, etc.), but the model and its parameters (such as temperature) are not automatically updated—they remain the ones used in the last session. Could OpenLumara be made to auto-detect the model upon startup and automatically update it along with its parameters?.
I'm very new to all this, and I don't know if this is the right place to ask for it; I apologize in advance!!!.

[Rose22]

1. good idea, i would need translation help though because i only speak 2 languages
2. what do you mean? like if it generates HTML inside a message, a button that lets you view that HTML as a rendered page?
3. this is fixed in the rewrite im doing. you no longer need to reconnect the api or restart the server for api/model settings to apply. but for parameters i recommend doing that on the side of your inference server (llamacpp, koboldcpp, etc), so that you dont have to keep swapping parameters in a UI. thats also the reason my ui doesn't have those parameters exposed.. i don't set them, so that your inference server can set them

[planetadelexilio-boop]

Thanks for replying and for all your work!!! Point "1": we're in the same boat. Regarding point "2", what I'm proposing is something similar to what the llama.cpp WebUI already does. I understand you mean the same thing :-). As for point "3", what confuses me is that I can modify the "Temp" parameter from OpenLumara, and I understand that this overrides my llama-server configuration. If I change the value to match my llama-server setup, that value doesn't change if I restart the session with a new LLM (with Qwen3.6 I use "0.6" for the temperature value, while with Gemma I use "1"). So the values ​​I set in OpenLumara become out of sync with those in llama-server when I switch LLMs.

[Rose22]

1. aw
2. i'll check it out so i can see what you mean (i currently use lemonade which wraps around llamacpp but stops the llamacpp webui from being served.. need to figure out how to make it show up)
3. oh, good point! yknow what, i'll just add a toggle "send sampling parameters", then add a few common ones, but default it to off. then with it off it wont send any overwrite values, but with it on you get control over stuff like top-k, top-p, etc