mainfile cgpt 8

app.py

@@ -25,14 +25,16 @@ llm = Together(
     model="mistralai/Mixtral-8x22B-Instruct-v0.1",
     temperature=0.2,
     top_k=12,
-    together_api_key=os.environ['pilotikval']
+    together_api_key=os.environ['pilotikval'],
+    max_tokens=200
 )
 
 llmc = Together(
     model="mistralai/Mixtral-8x22B-Instruct-v0.1",
     temperature=0.2,
     top_k=3,
-    together_api_key=os.environ['pilotikval']
+    together_api_key=os.environ['pilotikval'],
+    max_tokens=200
 )
 
 # Memory setup
@@ -67,15 +69,19 @@ def store_chat_history(role: str, content: str):
 
 # Define the chain using LCEL
 def create_conversational_qa_chain(retriever, condense_llm, answer_llm):
-    condense_question_chain = RunnableLambda(
-        lambda x: {"chat_history": chistory, "question": x['question']}
-    ) | CONDENSE_QUESTION_PROMPT | RunnableLambda(lambda x: {"standalone_question": x['standalone_question']})
+    condense_question_chain = (
+        RunnableLambda(lambda x: {"chat_history": chistory, "question": x['question']})
+        | CONDENSE_QUESTION_PROMPT
+        | RunnableLambda(lambda x: {"standalone_question": x})
+    )
 
-    retrieval_chain = RunnableLambda(
-        lambda x: {"standalone_question": x['standalone_question']}
-    ) | retriever | _combine_documents
+    retrieval_chain = (
+        RunnableLambda(lambda x: x['standalone_question'])
+        | retriever
+        | RunnableLambda(lambda x: {"context": _combine_documents(x)})
+    )
 
-    answer_chain = ANSWER_PROMPT | RunnableLambda(lambda x: {"context": x, "question": x['standalone_question']}) | answer_llm
+    answer_chain = ANSWER_PROMPT | answer_llm
 
     return RunnableParallel(
         condense_question=condense_question_chain,