Update app.py

app.py

@@ -19,15 +19,22 @@ from langchain.prompts import PromptTemplate
 from langchain.prompts.prompt import PromptTemplate
 from langchain.chat_models import ChatOpenAI
 
+LOAD_MODEL=False
+tokenizer,model = None,None
 
-tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True)
-model = AutoModel.from_pretrained("THUDM/chatglm-6b-int4", trust_remote_code=True).quantize(bits=4, compile_parallel_kernel=True, parallel_num=2).float()
-model = model.eval()
+def load_model():
+    tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True)
+    model = AutoModel.from_pretrained("THUDM/chatglm-6b-int4", trust_remote_code=True).quantize(bits=4, compile_parallel_kernel=True, parallel_num=2).float()
+    model = model.eval()
+    return tokenizer,model
 
 
 def chat_glm(input, history=None):
     if history is None:
         history = []
+    if not LOAD_MODEL:
+        LOAD_MODEL=True
+        tokenizer,model = load_model()
     response, history = model.chat(tokenizer, input, history)
     logger.info("chatglm:", input,response)
     return history, history
@@ -78,7 +85,7 @@ def chat_gpt(input, use_web, history=None):
         return_source_documents=True
     )
     
-    result = qa({"query": query, "chat_history": history})
+    result = qa({"query": input, "chat_history": history})
     logger.info("chatgpt:", input,result)
     return result["answer"]