André Pedersen

I would recommend using Zenodo. It is trivial to setup. Just link the GitHub acvount and make a release. You get an actual DOI and you ensure that the project...

Nothing stops you from having a DOI on the source code and a separate one for the publication. That is quite common and I have done the same as different...

> In the short term, I can tag a release (but see the caveat in the previous paragraph). But would you need to make a new tag for every single...

> If I ask the the LLM THESAME question in the question-sql pair that I loaded into chromadb, is there a way to get the LLM to use the corresponding...

> Yes you are right. You described the EXPECTED behavior, but the LLM still returns the wrong sql. Ive done several iterations but same result...hence the feature request... I believe...

@zainhoda I assume this feature was not part of the last release? Would be of great interest for our applications as well :]

@jankariwo Can you make a PR?

> By the way the original Sql-coder repo is also implemented using the Hugging Face Transformers library with its own prompt template. As pointed out in [#303 (comment)](https://github.com/vanna-ai/vanna/issues/303#issuecomment-2052692155), it would...

> I've used the 7b of sqlcoder via Ollama and found it to be extremely slow for some reason compared to models like mistral. > > I think if we...

This would be of interest to me as well. We are able to do this in Streamlit without any additional libraries. Does anyone know of a custom `gr.Chatbot` widget, which...