- Set your Ollama endpoint and model under Settings
- Upload your documents for processing
- Once complete, ask questions based on your documents!
All options within the RAG pipeline are exposed to users after toggling Settings > Show Advanced Options.
| Setting | Description | Default |
|---|---|---|
| Ollama Endpoint | The location of your locally hosted Ollama API | http://localhost:11434 |
| Model | Large language model to use what generating chat completions | |
| System Prompt | Initial system prompt used when initializing the LLM | (Please see source code) |
| Top K | Number of most similar documents to retrieve in response to a query | 3 |
| Chat Mode | Llama Index chat mode to utilize during retrievals | Best |
| Setting | Description | Default |
|---|---|---|
| Embedding Model | Embedding model to be used for vectorize your files | bge-large-en-v1.5 |
| Chunk Size | Improves embedding precision by focusing on smaller text portions | 1024 |