import onnxruntime
session = onnxruntime.InferenceSession("model.onnx")
print("Inputs:")
for idx, inputs in enumerate(session.get_inputs()):
print("idx:", idx)
print(" Name:", inputs.name)
print(" Shape:", inputs.shape)
print("Outputs:")
for idx, outputs in enumerate(session.get_outputs()):
print("idx:", idx)
print(" Name:", outputs.name)
print(" Shape:", outputs.shape)
outputs = session.run(None, {"input_ids": [[1, 2, 3]], "attention_mask": [[1, 1, 1]]})[0]
print(outputs.shape)