Update app.py

app.py

@@ -12,7 +12,7 @@ config.pre_seq_len = 128
 config.prefix_projection = False
 
 tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, trust_remote_code=True)
-model = AutoModel.from_pretrained(model_name_or_path, config=config, trust_remote_code=True)
+model = AutoModel.from_pretrained(model_name_or_path, config=config, trust_remote_code=True).float()
 
 # 加载 Prefix 参数
 prefix_state_dict = torch.load("pytorch_model.bin", map_location="cpu")
@@ -23,7 +23,6 @@ for k, v in prefix_state_dict.items():
 model.transformer.prefix_encoder.load_state_dict(new_prefix_state_dict)
 
 model = model.to(device)
-model = model.quantize(4)
 model.eval()
 
 # ------------------- Markdown 渲染 -------------------