SparseML create_pruning_recipe API erroring out on YOLOv3 tiny

[markurtz]

From community slack https://discuss-neuralmagic.slack.com/team/U03P31PUJGM:

Trunov Pavel 11:05 AM Hello! I'm trying to prune yolov3-tiny using sparseml. There is no recipe for tiny but I've figured out that it's possible to create a new one: link But I get error while converting: /usr/local/lib/python3.7/dist-packages/sparseml/onnx/sparsification/model_info.py in _get_model_last_prunable_nodes(model) 216 continue 217 for parent in graph.get_node_parents(node): --> 218 if any(output_id in seen_node_ids for output_id in parent.output): 219 continue 220 stack.append(parent)

AttributeError: output Could anyone help me? sparseml.version: 1.0.1 onnx.version: 1.10.1 code: from sparseml.sparsification import create_pruning_recipe ONNX_FILE_PATH = "yolov3_tiny.onnx" create_pruning_recipe(model=ONNX_FILE_PATH, save_path='./') model (netron): Inputs: name: images type: float32[1,3,416,416] Outputs: name: output type: float32[Concatoutput_dim_0,Concatoutput_dim_1,Concatoutput_dim_2] name: onnx::Sigmoid_124 type: float32[Transposeonnx::Sigmoid_124_dim_0,Transposeonnx::Sigmoid_124_dim_1,Transposeonnx::Sigmoid_124_dim_2,Transposeonnx::Sigmoid_124_dim_3,Transposeonnx::Sigmoid_124_dim_4] name: onnx::Sigmoid_190 type: float32[Transposeonnx::Sigmoid_190_dim_0,3,Transposeonnx::Sigmoid_190_dim_2,Transposeonnx::Sigmoid_190_dim_3,11] (edited)