Initial Commit

app.py

@@ -11,14 +11,18 @@ import math
 import logging
 import sys
 
-import os
-# 设置环境变量，强制禁用 accelerate 的显存管理
-os.environ["ACCELERATE_USE_MEMORY_EFFICIENT_ATTENTION"] = "false"
-os.environ["ACCELERATE_DISABLE_MEMORY_EFFICIENT_ATTENTION"] = "1"
-os.environ["PYTORCH_CUDA_ALLOC_CONF"] = "expandable_segments:True,garbage_collection_threshold:0.6,max_split_size_mb:512"
+from huggingface_hub.utils import HfFolder  # 添加这个导入
 
 from qwen2_vl.modeling_qwen2_vl import Qwen2VLSimplifiedModel
 
+# 在代码最开始添加 ZeroGPU 初始化
+try:
+    from accelerate.utils import ZeroGPU
+    zerogpu = ZeroGPU()
+    zerogpu.init()
+except ImportError:
+    print("Warning: ZeroGPU not available, falling back to CPU")
+
 # 设置日志
 logging.basicConfig(
     level=logging.INFO,
@@ -440,5 +444,6 @@ if __name__ == "__main__":
     demo.launch(
         server_name="0.0.0.0",  # Listen on all network interfaces
         server_port=7860,       # Use a specific port
-        share=False             # Disable public URL sharing
+        share=False,             # Disable public URL sharing
+        enable_queue=True  # 添加队列支持
     )