Update app.py

app.py

@@ -12,16 +12,15 @@ For more information on `huggingface_hub` Inference API support, please check th
 """
 
 model_name = "llama-3.3-70b-versatile"
-embeddings = HuggingFaceEmbeddings(
-    model_name = "pkshatech/GLuCoSE-base-ja"
-)
-vector_store = InMemoryVectorStore.load(
-    "kaihatsu_vector_store", embeddings
-)
-retriever = vector_store.as_retriever(search_kwargs={"k": 4})
+emb_model_name = "pkshatech/GLuCoSE-base-ja"
 
+def load_vector_store(embedding_model_name, vector_store_file, k=4):
+    embeddings = HuggingFaceEmbeddings(model_name = embedding_model_name)
+    vector_store = InMemoryVectorStore.load(vector_store_file, embeddings)
+    retriever = vector_store.as_retriever(search_kwargs={"k": k})
+    return retriever
 
-def fetch_response(groq_api_key, user_input):
+def fetch_response(groq_api_key, user_input, retriever1, retriever2):
   chat = ChatGroq(
     api_key = groq_api_key,
     model_name = model_name
@@ -42,15 +41,20 @@ def fetch_response(groq_api_key, user_input):
   # ドキュメントのリストを渡せるchainを作成
   question_answer_chain = create_stuff_documents_chain(chat, prompt)
   # RetrieverとQAチェーンを組み合わせてRAGチェーンを作成
-  rag_chain = create_retrieval_chain(retriever, question_answer_chain)
+  rag_chain = create_retrieval_chain(retriever1, question_answer_chain)
+  rag_chain2 = create_retrieval_chain(retriever2, question_answer_chain)
 
   response = rag_chain.invoke({"input": user_input})
-  return [response["answer"], response["context"][0], response["context"][1]]
+  response2 = rag_chain.invoke({"input": user_input})
+  return [response["answer"], response["context"][0], response["context"][1], response2["answer"], response2["context"][0], response2["context"][1]]
 
 
 """
 For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
 """
+retriever = load_vector_store(emb_model_name, "kaihatsu_vector_store", 4)
+retriever_rechunk = load_vector_store(emb_model_name, "kaihatsu_vector_store_rechunk", 4)
+
 with gr.Blocks() as demo:
     gr.Markdown('''# 「スマート農業技術の開発・供給に関する事業」マスター \n
                      「スマート農業技術の開発・供給に関する事業」に関して、公募要領や審査要領を参考にRAGを使って回答します。
@@ -62,12 +66,16 @@ with gr.Blocks() as demo:
             user_input = gr.Textbox(label="User Input")
             submit = gr.Button("Submit")
             answer = gr.Textbox(label="Answer")
+            answer2 = gr.Textbox(label="Answer")
     with gr.Row():
         with gr.Column():
             source1 = gr.Textbox(label="回答ソース１")
+            source2_1 = gr.Textbox(label="回答ソース１")
         with gr.Column():
             source2 = gr.Textbox(label="回答ソース２")
-    submit.click(fetch_response, inputs=[api_key, user_input], outputs=[answer, source1, source2])
+            source2_2 = gr.Textbox(label="回答ソース２")
+            
+    submit.click(fetch_response, inputs=[api_key, user_input, retriever, retriever_rechunk], outputs=[answer, source1, source2, answer2, source2_1, source2_2])
 
 if __name__ == "__main__":
     demo.launch()