Update app.py

app.py

@@ -842,30 +842,50 @@ questions = ["what does DTC stands for?",
 
 # Instantiate the retriever and the ConversationalRetrievalChain :
 
+
 retriever_HF = retrieval_blocks(
     create_vectorstore=False,
-    LLM_service="Huggingface",
+    LLM_service="HuggingFace",
     vectorstore_name="Vit_All_HF_Embeddings",
     retriever_type="Cohere_reranker",
-    base_retriever_search_type="similarity", base_retriever_k=12,
-    compression_retriever_k=16,
+    base_retriever_search_type="similarity", base_retriever_k=16,
+    compression_retriever_k=20,
     cohere_api_key=cohere_api_key,cohere_top_n=10,
 )
 
-chain_gemini,memory_gemini = ConversationalRetrievalChain(
+chain_HF,memory_HF = custom_ConversationalRetrievalChain(
     llm = instantiate_LLM(
-        LLM_provider="Google",api_key=google_api_key,temperature=0.5,model_name="gemini-pro"
+        LLM_provider="HuggingFace",api_key=HF_key,temperature=0.5,
+        model_name="mistralai/Mistral-7B-Instruct-v0.2"
     ),
     condense_question_llm = instantiate_LLM(
-        LLM_provider="Google",api_key=google_api_key,temperature=0.1,model_name="gemini-pro"),
-    retriever=base_retriever_HF,
+        LLM_provider="HuggingFace",api_key=HF_key,temperature=0.5,
+        model_name="mistralai/Mistral-7B-Instruct-v0.2"
+    ),
+    retriever=retriever_HF,
     language="english",
-    llm_provider="Google",
-    model_name="gemini-pro"
+    llm_provider="HuggingFace",
+    model_name="Mistral-7B-Instruct-v0.2"
 )
 
 
-memory_gemini.clear()
+memory_HF.clear()
+
+response = chain_HF.invoke({"question":questions[0]})
+    
+answer = response['answer']
+answer = answer[answer.find("\nAnswer: ")+len("\nAnswer: "):]
+
+gr.markdown("**Question:** "+questions[0]+"\n\n"\
++"**Standalone_question:**"+response['standalone_question']+"\n\n"\
++"**Answer:** "+answer)
+
+
+
+
+
+
+
 
 for i,question in enumerate(questions):
     response = chain_gemini.invoke({"question":question})