Update app.py

app.py

@@ -61,9 +61,7 @@ def qwen_api(user_message, top_p=0.9,temperature=0.7, system_message='', max_tok
 
 os.environ["OPENAI_API_BASE"] = "https://api-inference.huggingface.co/v1/"
 os.environ["OPENAI_API_KEY"] = TOKEN
-llm = ChatOpenAI(
-        model="meta-llama/Meta-Llama-3-8B-Instruct",
-        temperature=0.8,)
+
 
 
 
@@ -79,13 +77,22 @@ PROMPT = PromptTemplate(
 )
 chain_type_kwargs = {"prompt": PROMPT}
 retriever = db.as_retriever()
-qa = RetrievalQA.from_chain_type(
-        llm=llm,
-        chain_type="stuff",
-        retriever=retriever,
-        chain_type_kwargs=chain_type_kwargs,
-        return_source_documents=True
-    )
+
+def langchain_chat(message, temperature, top_p, max_tokens):
+    llm = ChatOpenAI(
+        model="meta-llama/Meta-Llama-3-8B-Instruct",
+        temperature=temperature,
+        top_p=top_p,
+        max_tokens=max_tokens)
+    qa = RetrievalQA.from_chain_type(
+            llm=llm,
+            chain_type="stuff",
+            retriever=retriever,
+            chain_type_kwargs=chain_type_kwargs,
+            return_source_documents=True
+        )
+    response = qa.invoke(message)['result']
+    return response
 
 def chat(
     message,
@@ -96,9 +103,9 @@ def chat(
     top_p,
 ):
     if len(history) == 0:
-        response = qa.invoke(message)['result']
+        response = langchain_chat(message, temperature, top_p, max_tokens)
     else:
-        response = qwen_api(message, gradio_history=history)
+        response = qwen_api(message, gradio_history=history, max_tokens=max_tokens, top_p=top_p, temperature=temperature)
     print(response)
     yield response
     return response