LLM + RAG to generate answers about the manifestos of the parties for the 2024 Austrian legislative election.
as described in the blog post: https://johnnys.news/2024/09/How-I-built-wahlGPT
- .NET SDK >= 8
- wasm workload
- Azure Functions Core Tools
# run the backend function
cd WahlGPT.Function
func start
# adapt settings (i.e. API_ENDPOINT from function) in WahlGPT.Common/Settings.cs
# for development the following extensions for KernelMemory can be used:
#.WithSimpleVectorDb()
#.WithSimpleFileStorage()
# with these you don't need a running qdrant instance - the vector db is stored in memory
# feel free to plug in other models for embedding or text generation to experiment with different results
#restore the wasm workloads
cd WahlGPT.Web
dotnet workload restore
# run the frontend
dotnet run- html5up
- semantic-kernel
- semantic-memory
- qdrant
- dotnet
- openai