Customized yolov8-pose model, and the number of key points mask data is not equal to 56

[ccl-private]

I have trained a custom key-points model, which causes 'mask' data to be not 56. so I modified two places. First assert classes_masks_out == custom_num; Second, post_processing_steps Split(num_outputs=3, axis=1, splits=[4, 1, custom_num]).
But when running it, it reported an error. Here is the output log:
Updating model with jpg/png image bytes as input.
Adding pre/post processing to the model...
Updating model ...
Pre/post proceessing added.
Updated model saved.
Running the model to validate output using /slow_disk/ccl/data/goji_pose/data/2024-10-14 095626.jpg.
Traceback (most recent call last):
File "/slow_disk/ccl/codes/yolo_custom/onnx_core_goji.py", line 373, in
_run_inference(updated_model_path, args.input, args.output_image is not None, args.test_image, rgb_layout)
File "/slow_disk/ccl/codes/yolo_custom/onnx_core_goji.py", line 240, in _run_inference
session = ort.InferenceSession(str(onnx_model_path), providers=providers, sess_options=session_options)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/slow_disk/ccl/codes/yolo_custom/.venv/lib/python3.11/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 419, in init
self._create_inference_session(providers, provider_options, disabled_optimizers)
File "/slow_disk/ccl/codes/yolo_custom/.venv/lib/python3.11/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 491, in _create_inference_session
sess.initialize_session(providers, provider_options, disabled_optimizers)
onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Node (post_process_57) Op (Reshape) [ShapeInferenceError] Invalid position of 0.