chunked/nltk_chunking/_serialization/chunk_5.txt

When a model is exported to the ONNX format, these operators are used to
construct a computational graph (often called an intermediate representation) which
represents the flow of data through the neural network.