Update handler.py

handler.py

@@ -13,8 +13,10 @@ class EndpointHandler():
     def __init__(self, path=""):
         self.device = "cuda:0" if torch.cuda.is_available() else "cpu"
         print("device:",self.device)
-        self.model_base = "Salesforce/blip2-opt-2.7b"
-        self.model_name = "sooh-j/blip2-vizwizqa"
+        # self.model_base = "Salesforce/blip2-opt-2.7b"
+        # self.model_name = "sooh-j/blip2-vizwizqa"
+        self.model_name = "Salesforce/blip2-opt-2.7b"
+
         self.processor = AutoProcessor.from_pretrained(self.model_name)
         self.model = Blip2ForConditionalGeneration.from_pretrained(self.model_name,
                                                               device_map="auto", 
@@ -50,7 +52,7 @@ class EndpointHandler():
 
         with torch.no_grad():
             out = self.model.generate(**processed, 
-                                     max_new_tokens=50,
+                                     max_new_tokens=20,
                                      temperature = 0.5,
                                      do_sample=True,
                                      top_k=50,