PaddlePaddle has introduced HPD-Parsing, a lightweight 1-billion parameter document parsing model that utilizes a Hierarchical Parallel Decoding paradigm to improve throughput. The architecture coordinates global page structure through a main layout branch while dispatching localized content generation to concurrent branches, utilizing Progressive Multi-Token Prediction to reduce decoding steps.

  • HPD-Parsing achieves an overall score of 94.91% on OmniDocBench v1.6, setting a new state of the art among end-to-end unified parsers.
  • The model reaches a peak throughput of 4,752 Tokens Per Second (TPS), which is 3.06 times faster than its autoregressive baseline and 2.62 times faster than the fastest existing document parser.
  • A staged adaptation strategy with automated difficulty-aware data curation preserves parsing accuracy during the transition to parallel decoding.

This approach demonstrates that document parsing can be effectively executed through global layout coordination and localized parallel decoding rather than a single sequential generation trajectory.