Update QnA.py

QnA.py

@@ -54,17 +54,28 @@ def summarize(documents,llm):
     return results['output_text']
 
 
+def get_hugging_face_model(model_id='mistralai/Mistral-7B-Instruct-v0.3',temperature=0.01,max_tokens=2048,api_key=None):
+    llm = HuggingFaceHub(
+        huggingfacehub_api_token =api_key ,
+        repo_id=model_id, 
+        model_kwargs={"temperature":temperature, "max_new_tokens":max_tokens}
+        )
+    return llm
+
 def Q_A(vectorstore,question,API_KEY):
-    os.environ["GROQ_API_KEY"] = API_KEY
-    llm_groq = ChatGroq(model="llama3-8b-8192")
+    if API_KEY.startswith('gsk'):
+        os.environ["GROQ_API_KEY"] = API_KEY
+        chat_llm = ChatGroq(model="llama3-8b-8192")
+    elif API.startswith('hk'):
+        chat_llm = get_hugging_face_model(API_key=API_KEY)
     
     # Create a retriever
     retriever = vectorstore.as_retriever(search_type = 'similarity',search_kwargs = {'k':2},)
     if  'reliable' in question.lower() or 'relaibility' in question.lower():
-        question_answer_chain = create_stuff_documents_chain(llm_groq, prompt_template_for_relaibility())
+        question_answer_chain = create_stuff_documents_chain(chat_llm, prompt_template_for_relaibility())
         
     else:
-        question_answer_chain = create_stuff_documents_chain(llm_groq, prompt_template_to_analyze_resume())
+        question_answer_chain = create_stuff_documents_chain(chat_llm, prompt_template_to_analyze_resume())
     
     chain = create_retrieval_chain(retriever, question_answer_chain)
     result = chain.invoke({'input':question})