Use fine tuned embedding model in Gradio UI

README.md

@@ -82,8 +82,7 @@ Agentic reasoning allows contextual memory, thematic consistency, and accurate m
 
 | Source / API                                          | Purpose                                                                                                                                                                                         |
 | ----------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
-| **`kjv_preprocessed.csv`**                            | Core knowledge base of biblical content. Each row includes `Text`, `Book`, `Chapter`, and `Verse`, allowing semantic indexing and metadata filtering.                                           |
-| **Tavily Search API**                                 | For agent-initiated internet augmentation (e.g., cross-referencing external theological commentaries, definitions, or historical context). Used as a fallback for missing or ambiguous context. |
+| **`kjv.csv`**                            | Core knowledge base of biblical content. Each row includes `Text`, `Book`, `Chapter`, and `Verse`, allowing semantic indexing and metadata filtering.                                           |
 | **OpenAI APIs (`gpt-4.1`, `text-embedding-3-small`)** | Language generation and semantic embedding for both retrieval and agent reasoning.                                                                                                              |
 | **LangSmith**                                         | Observability and debugging of multi-agent reasoning chains.                                                                                                                                    |
 | **RAGAS**                                             | Evaluation of the RAG pipeline using metrics like **Faithfulness**, **Answer Relevance**, **Context Precision**, and **Context Recall**.                                                        |

gradio_app.py

@@ -87,7 +87,7 @@ with gr.Blocks(theme=gr.themes.Soft(), css="footer {display:none !important}") a
 
     vectorstore = create_qdrant_vectorstore(
         documents=chunked_docs_df.to_dict("records"),
-        embedding_model=get_openai_embedding_model(),
+        embedding_model=get_finetuned_embedding_model(),
         collection_name=settings.COLLECTION_NAME_FINETUNED,
     )