ZeroGPU fix attempt

model/backend.py

@@ -15,12 +15,13 @@ import os, json, re, urllib.request
 
 # ZeroGPU: decorate GPU-using functions with @_GPU so HF Spaces can allocate
 # the GPU on demand. Falls back to a no-op when running locally (no `spaces` pkg).
-try:
-    import spaces
-    _GPU = spaces.GPU
-except ImportError:
-    def _GPU(fn=None, *, duration=60):
-        return fn if fn is not None else (lambda f: f)
+import spaces
+# try:
+#     import spaces
+#     # _GPU = spaces.GPU
+# except ImportError:
+#     def _GPU(fn=None, *, duration=60):
+#         return fn if fn is not None else (lambda f: f)
 
 # is_torch_fx_available fix
 import transformers.utils.import_utils
@@ -188,7 +189,8 @@ def _get_hf_pipe():
 # Stateless (one-turn only), strictly prevents thinking tokens per Qwen3 docs.
 _NO_THINK_SUFFIX = [{"role": "assistant", "content": "<think>\n\n</think>\n\n"}]
 
-@_GPU(duration=120)
+# @_GPU(duration=120)
+@spaces.GPU(duration=120)
 def _hf_generate(messages, max_new_tokens=512, temperature=0.6) -> str:
     pipe = _get_hf_pipe()
     full_messages = messages + _NO_THINK_SUFFIX