Update main.py

main.py

@@ -2,6 +2,7 @@
 from fastapi import FastAPI, File, UploadFile
 import os
 import shutil
+from LoadLLM import Loadllm
 import tempfile
 from langchain_community.document_loaders import PyMuPDFLoader
 from langchain_community.llms import CTransformers
@@ -37,12 +38,13 @@ async def PromptLLM(file: UploadFile = File(...)):
     db.save_local(DB_FAISS_PATH)
 
     # Load the language model
-    llm = CTransformers(model="llama-2-7b-chat.ggmlv3.q4_1.bin", model_type="llama",config={'max_new_tokens': 1024,  'context_length': 2048, 'temperature': 0.01})
+    # Load the language model
+    llm = Loadllm.load_llm()
 
     # Create a conversational chain
     chain = ConversationalRetrievalChain.from_llm(llm=llm, retriever=db.as_retriever())
 
-    result = chain({"question": "Summarise the report", "chat_history": ''})
+    result = chain({"question": query, "chat_history": ''})
     return result['answer']