GODEL: Large-Scale Pre-Training for Goal-Directed Dialog
This work addresses the challenge of building more useful and grounded dialog systems for applications like task-oriented assistance, though it is incremental in advancing pre-training techniques.
The authors tackled the problem of adapting large pre-trained language models to goal-directed dialog tasks requiring external information by introducing GODEL, which outperformed state-of-the-art models in few-shot fine-tuning across multiple benchmarks, achieving improvements in both human and automatic evaluations.
We introduce GODEL (Grounded Open Dialogue Language Model), a large pre-trained language model for dialog. In contrast with earlier models such as DialoGPT, GODEL leverages a new phase of grounded pre-training designed to better support adapting GODEL to a wide range of downstream dialog tasks that require information external to the current conversation (e.g., a database or document) to produce good responses. Experiments against an array of benchmarks that encompass task-oriented dialog, conversational QA, and grounded open-domain dialog show that GODEL outperforms state-of-the-art pre-trained dialog models in few-shot fine-tuning setups, in terms of both human and automatic evaluation. A novel feature of our evaluation methodology is the introduction of a notion of utility that assesses the usefulness of responses (extrinsic evaluation) in addition to their communicative features (intrinsic evaluation). We show that extrinsic evaluation offers improved inter-annotator agreement and correlation with automated metrics. Code and data processing scripts are publicly available.