Lattice-BERT: Leveraging Multi-Granularity Representations in Chinese Pre-trained Language Models
This work addresses a domain-specific bottleneck in Chinese NLP by improving model performance through multi-granularity processing, though it is incremental as it builds on existing transformer-based pre-training paradigms.
The authors tackled the problem of Chinese pre-trained language models ignoring word-level granularity by proposing Lattice-BERT, a model that incorporates multi-granularity representations, resulting in an average increase of 1.5% on 11 Chinese NLU tasks and achieving state-of-the-art performance among base-size models on CLUE benchmarks.
Chinese pre-trained language models usually process text as a sequence of characters, while ignoring more coarse granularity, e.g., words. In this work, we propose a novel pre-training paradigm for Chinese -- Lattice-BERT, which explicitly incorporates word representations along with characters, thus can model a sentence in a multi-granularity manner. Specifically, we construct a lattice graph from the characters and words in a sentence and feed all these text units into transformers. We design a lattice position attention mechanism to exploit the lattice structures in self-attention layers. We further propose a masked segment prediction task to push the model to learn from rich but redundant information inherent in lattices, while avoiding learning unexpected tricks. Experiments on 11 Chinese natural language understanding tasks show that our model can bring an average increase of 1.5% under the 12-layer setting, which achieves new state-of-the-art among base-size models on the CLUE benchmarks. Further analysis shows that Lattice-BERT can harness the lattice structures, and the improvement comes from the exploration of redundant information and multi-granularity representations. Our code will be available at https://github.com/alibaba/pretrained-language-models/LatticeBERT.