Merge pull request #71 from HagaiAstrin/RAG_Techniques_Hagai

Fix for RuntimeWarning: invalid value encountered in divide.
This commit is contained in:
NirDiamant
2024-12-25 18:10:32 +02:00
committed by GitHub

View File

@@ -247,6 +247,9 @@
" Returns:\n",
" List[Document]: The top k documents based on the combined scores.\n",
" \"\"\"\n",
" \n",
" epsilon = 1e-8\n",
"\n",
" # Step 1: Get all documents from the vectorstore\n",
" all_docs = vectorstore.similarity_search(\"\", k=vectorstore.index.ntotal)\n",
"\n",
@@ -258,9 +261,9 @@
" \n",
" # Step 4: Normalize scores\n",
" vector_scores = np.array([score for _, score in vector_results])\n",
" vector_scores = 1 - (vector_scores - np.min(vector_scores)) / (np.max(vector_scores) - np.min(vector_scores))\n",
" vector_scores = 1 - (vector_scores - np.min(vector_scores)) / (np.max(vector_scores) - np.min(vector_scores) + epsilon)\n",
"\n",
" bm25_scores = (bm25_scores - np.min(bm25_scores)) / (np.max(bm25_scores) - np.min(bm25_scores))\n",
" bm25_scores = (bm25_scores - np.min(bm25_scores)) / (np.max(bm25_scores) - np.min(bm25_scores) + epsilon)\n",
"\n",
" # Step 5: Combine scores\n",
" combined_scores = alpha * vector_scores + (1 - alpha) * bm25_scores \n",