diff --git a/include/core/tensor.h b/include/core/tensor.h
index 1edc950a..c5823be0 100644
--- a/include/core/tensor.h
+++ b/include/core/tensor.h
@@ -32,7 +32,7 @@ class TensorObj : public TensorBaseObj {
     using TensorBaseObj::getData;
     VType getData(const Shape &pos) const;
     void dataMalloc();
-    UidBaseType getFuid() const { return fuid; }
+    inline UidBaseType getFuid() const { return fuid; }
 
     void load(std::string file_path);
     void save(std::string file_path);
diff --git a/pyinfinitensor/src/pyinfinitensor/onnx.py b/pyinfinitensor/src/pyinfinitensor/onnx.py
index 28b8f514..d554883c 100644
--- a/pyinfinitensor/src/pyinfinitensor/onnx.py
+++ b/pyinfinitensor/src/pyinfinitensor/onnx.py
@@ -35,6 +35,7 @@ def cuda_runtime():
 class OnnxStub:
     inputs: Dict[str, backend.Tensor] = {}
     outputs: Dict[str, backend.Tensor] = {}
+    initializer: Dict[int, TensorProto] = {}
     handler: backend.GraphHandler
 
     def __init__(self, model: ModelProto, runtime):
@@ -348,6 +349,7 @@ class OnnxStub:
                 if any(input.name == name for input in model.graph.input):
                     self.inputs[name] = obj
             else:
+                self.initializer[obj.fuid()] = tensor
                 if tensor.data_type == TensorProto.INT32:
                     self.handler.copy_int32(obj, [int(i) for i in tensor.int32_data])
                 elif tensor.data_type == TensorProto.INT64:
diff --git a/src/ffi/ffi_infinitensor.cc b/src/ffi/ffi_infinitensor.cc
index 87599f28..d7230f42 100644
--- a/src/ffi/ffi_infinitensor.cc
+++ b/src/ffi/ffi_infinitensor.cc
@@ -154,6 +154,7 @@ void init_graph_builder(py::module &m) {
         m, "CudaRuntime");
 #endif
     py::class_<TensorObj, std::shared_ptr<TensorObj>>(m, "Tensor")
+        .def("fuid", &TensorObj::getFuid, policy::automatic)
         .def("shape", &TensorObj::getDims, policy::move)
         .def("cloneFloats", &TensorObj::cloneFloats, policy::move)
         .def("has_target", &TensorObj::hasTarget, policy::automatic)