Update app.py

app.py

@@ -184,8 +184,9 @@ def load_model():
     # model = (AutoModelForCausalLM.from_pretrained(
     #     model_name_or_path,
     #     trust_remote_code=True).to(torch.bfloat16).cuda())
-    model = AutoModelForCausalLM.from_pretrained(model_name_or_path, map_location=torch.device('cpu'),trust_remote_code=True)
-    model.to_bettertransformer()
+    model = AutoModelForCausalLM.from_pretrained(
+        model_name_or_path,
+        trust_remote_code=True).to(torch.float32)
     tokenizer = AutoTokenizer.from_pretrained(model_name_or_path,
                                               trust_remote_code=True)
     return model, tokenizer