CLMay 13, 2024

DEPTH: Discourse Education through Pre-Training Hierarchically

arXiv:2405.07788v212 citationsh-index: 55Has CodeRepL4NLP
Originality Incremental advance
AI Analysis

This addresses discourse understanding limitations in language models, which is important for NLP applications requiring coherence and narrative flow, though it appears incremental as it extends T5's capabilities.

The paper tackles the problem of language models struggling with discourse-level understanding by introducing DEPTH, an encoder-decoder model with hierarchical sentence representations and discourse-oriented pre-training objectives. The result shows DEPTH learns semantic and discourse representations faster than T5 and outperforms it in span-corruption loss while maintaining other NLU capabilities.

Language Models (LMs) struggle with linguistic understanding at the discourse level, even though discourse patterns such as coherence, cohesion, and narrative flow are prevalent in their pre-training data. To improve the discourse capabilities of LMs already at the pre-training stage, we introduce DEPTH, an encoder-decoder model that learns latent representations for sentences using a discourse-oriented pre-training objective. DEPTH combines hierarchical sentence representations with two objectives: (1) Sentence Un-Shuffling, and (2) Span-Corruption. Our approach trains the model to represent both sub-word-level and sentence-level dependencies over a pre-training corpora. When trained either from scratch or continuing from a pre-trained T5 checkpoint, DEPTH learns semantic and discourse-level representations faster than T5, outperforming it in span-corruption loss despite the additional sentence-un-shuffling objective. Evaluations on the GLUE, DiscoEval, and NI benchmarks demonstrate DEPTH's ability to quickly learn diverse downstream tasks, which require syntactic, semantic, and discourse capabilities. Our approach extends the discourse capabilities of T5, while minimally impacting other natural language understanding (NLU) capabilities in the resulting LM. We share our codebase for reproducibility: https://github.com/zbambergerNLP/depth.git.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes