Nils Herzig
Nils Herzig
I'm open to merge a working solution, but we have to keep in mind that every bit of extra prompt complexity will degrade the performance of small LLMs.
https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2 Looks like the embeddings model was only trained on English text? Maybe changing this would also improve the results
 mistral results using "nomic-embed-text" instead of "minilm". Results are good btw haha, i forgot that not everyone can judge that
Damn I'm going to switch the whole project to nomic, it's a bit larger but performs way better haha.
 looks good to me, right? using the default model (hermes 2 pro mistral 7b)
> I could not find info about Nomic model being multilingual on their technical info but... if it works, it works. The fact that it works will Ollama is also...
just exposed the system prompt and a lot of other settings to the frontend, might improve testing