Update app.py

app.py

@@ -1,48 +1,41 @@
 import os
 import torch
 import gradio as gr
-from transformers import pipeline, AutoModelForCausalLM, AutoTokenizer
-from typing import List, Tuple, Dict
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-# 模型配置 - 全部使用无访问限制的公开模型
+# 从环境变量获取Hugging Face Token
+hf_token = os.environ.get("HUGGINGFACE_HUB_TOKEN")
+if not hf_token:
+    raise EnvironmentError("未找到HUGGINGFACE_HUB_TOKEN环境变量，请在Space设置中添加")
+
+# 模型配置 - 使用公开模型
 MODELS = {
     "Zephyr 7B Beta": {
         "model_id": "HuggingFaceH4/zephyr-7b-beta",
         "kwargs": {"torch_dtype": torch.float16}
     },
-    "Mistral 7B Instruct": {
-        "model_id": "mistralai/Mistral-7B-Instruct-v0.2",
-        "kwargs": {"torch_dtype": torch.float16}
-    },
-    "OpenHermes 2.5": {
-        "model_id": "teknium/OpenHermes-2.5-Mistral-7B",
-        "kwargs": {"torch_dtype": torch.float16}
-    },
     "Falcon 7B Instruct": {
         "model_id": "tiiuae/falcon-7b-instruct",
-        "kwargs": {"torch_dtype": torch.float16}
+        "kwargs": {"torch_dtype": torch.float16, "trust_remote_code": True}
     }
 }
 
-# 加载模型和分词器
+# 加载模型
 def load_model(model_name):
     model_config = MODELS[model_name]
-    tokenizer = AutoTokenizer.from_pretrained(model_config["model_id"])
-    
-    # 处理特殊模型参数（如需要）
-    if "Falcon" in model_name:
-        model_config["kwargs"]["trust_remote_code"] = True
-    
+    tokenizer = AutoTokenizer.from_pretrained(
+        model_config["model_id"],
+        use_auth_token=hf_token
+    )
     model = AutoModelForCausalLM.from_pretrained(
         model_config["model_id"],
+        use_auth_token=hf_token,
         **model_config["kwargs"]
     )
-    
-    # 移动到可用设备
     device = "cuda" if torch.cuda.is_available() else "cpu"
-    model = model.to(device)
-    
-    return model, tokenizer, device
+    return model.to(device), tokenizer, device
+
+# 其余代码（界面构建和交互逻辑）保持不变...
 
 # 初始化模型
 loaded_models = {}