Update app.py

app.py

@@ -16,7 +16,8 @@ from langchain.vectorstores import Chroma
 
 ### 3. For Querying LLM
 # for loading HuggingFace LLM models from the hub
-from langchain.llms import HuggingFaceHub
+#from langchain.llms import HuggingFaceHub
+from langchain_community.llms import HuggingFaceEndpoint
 # for querying LLM conveniently using the context
 from langchain.chains.question_answering import load_qa_chain
 
@@ -42,8 +43,13 @@ chroma_db = Chroma(persist_directory='chromadb_earnings_transcripts_extracted/ch
 model = "mistralai/Mistral-7B-Instruct-v0.1"
 
 # This is an inference endpoint API from huggingface, the model is not run locally, it is run on huggingface
-hf_llm = HuggingFaceHub(repo_id=model,model_kwargs={'temperature':0.5,"max_new_tokens":300})
-
+# hf_llm = HuggingFaceHub(repo_id=model,model_kwargs={'temperature':0.5,"max_new_tokens":300})
+hf_llm = HuggingFaceEndpoint(
+    endpoint_url=model,
+    huggingfacehub_api_token=os.environ['HUGGINGFACEHUB_API_TOKEN'],
+    task="text-generation",
+    max_new_tokens=512
+)
 
 def source_question_answer(query:str,vectorstore:Chroma=chroma_db,llm:HuggingFaceHub=hf_llm):
   """