CoCoLex: Confidence-guided Copy-based Decoding for Grounded Legal Text Generation
This addresses the issue of unreliable text generation for legal professionals, but it is incremental as it builds on existing context-aware decoding approaches.
The paper tackled the problem of unfaithful or hallucinatory text generation by LLMs in the legal domain by introducing CoCoLex, a decoding strategy that dynamically interpolates model-generated distributions with context-based copying to improve fidelity, and it outperformed existing methods on five legal benchmarks, especially in long-form tasks.
Due to their ability to process long and complex contexts, LLMs can offer key benefits to the Legal domain, but their adoption has been hindered by their tendency to generate unfaithful, ungrounded, or hallucinatory outputs. While Retrieval-Augmented Generation offers a promising solution by grounding generations in external knowledge, it offers no guarantee that the provided context will be effectively integrated. To address this, context-aware decoding strategies have been proposed to amplify the influence of relevant context, but they usually do not explicitly enforce faithfulness to the context. In this work, we introduce Confidence-guided Copy-based Decoding for Legal Text Generation (CoCoLex)-a decoding strategy that dynamically interpolates the model produced vocabulary distribution with a distribution derived based on copying from the context. CoCoLex encourages direct copying based on the model's confidence, ensuring greater fidelity to the source. Experimental results on five legal benchmarks demonstrate that CoCoLex outperforms existing context-aware decoding methods, particularly in long-form generation tasks.