We made a web interface with the python streamlit library.
streamlit run app.py --server.address 0.0.0.0 --server.port portport is the port to expose
Meteo France
On priam the gpus are not powerful enough to run the generative models. So the script must be launched from the ewc. To access the interface from a browser you have to make an ssh tunnel.
on a terminal from priam do:
ssh -g -L port:0.0.0.0:port ewcport is the port to expose
- Number of retrieved chunks : choose the number of retrieved chunks that will be given to the generative model
- Use rag : enable or disable the use of rag
- Generative model : Select the desired generative model from the list. This corresponds to the model that will generate the final response.
- Reranker model : Select the desired reranker model from the list. This corresponds to the model that will select the most relevant chunks.