Mayank Chugh
Refactor embedding function creation and document loading. Update ingest and query routes to remove unnecessary settings parameters, streamline chunking logic, and enhance load_documents function to handle both string and list inputs. Adjust model name in embedder for consistency with OpenAI API.
830947a | Doc-Audi-AI RAG Smoke Test Document | |
| Project: Doc-Audi-AI | |
| Environment: Lightning AI deployment with Ollama embeddings. | |
| This sample document is used to test ingestion and retrieval. | |
| The system should split this file into chunks, generate embeddings, and store vectors in Chroma. | |
| Key facts: | |
| - The project supports file ingestion for PDF, TXT, and MD formats. | |
| - The default collection name for tests is "default". | |
| - A typical retrieval question is: "What is this document about?" | |
| - Another test question is: "Which file formats are supported?" | |
| Expected behavior: | |
| If ingestion succeeds, querying should return text snippets from this document with relevance scores. | |