Kris Cao


Control Prefixes for Parameter-Efficient Text Generation
Jordan Clive | Kris Cao | Marek Rei
Proceedings of the 2nd Workshop on Natural Language Generation, Evaluation, and Metrics (GEM)

Prefix-tuning is a parameter-efficient and powerful technique for adapting a pre-trained language model to a downstream application. However, it uses the same dataset-level tuned set of parameters for all examples in the dataset. We extend the framework with a dynamic method, Control Prefixes, which allows for the effective inclusion of input-dependent information, thereby demonstrating how prefix-tuning can be used for controlled text generation tasks. The method incorporates attribute-level learnable representations into different layers of a pre-trained Transformer, enabling the generated text to be guided in a particular direction. We provide a systematic evaluation of the technique and apply it to five datasets from the GEM benchmark for natural language generation (NLG). Using only 0.1–2% additional trainable parameters, we show Control Prefixes can even outperform full fine-tuning methods, and present state-of-the-art results on several data-to-text datasets, including WebNLG. We also examine the common case where input-dependent information is unavailable at test time and show Control Prefixes can excel in this setting also.


You should evaluate your language model on marginal likelihood over tokenisations
Kris Cao | Laura Rimell
Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing

Neural language models typically tokenise input text into sub-word units to achieve an open vocabulary. The standard approach is to use a single canonical tokenisation at both train and test time. We suggest that this approach is unsatisfactory and may bottleneck our evaluation of language model performance. Using only the one-best tokenisation ignores tokeniser uncertainty over alternative tokenisations, which may hurt model out-of-domain performance. In this paper, we argue that instead, language models should be evaluated on their marginal likelihood over tokenisations. We compare different estimators for the marginal likelihood based on sampling, and show that it is feasible to estimate the marginal likelihood with a manageable number of samples. We then evaluate a pretrained language model on both the one-best-tokenisation and marginal perplexities, and show that the marginal perplexity can be significantly better than the one best, especially on out-of-domain data. We link this difference in perplexity to the tokeniser uncertainty as measured by tokeniser entropy. We discuss some implications of our results for language model training and evaluation, particularly with regard to tokenisation robustness.


Factorising AMR generation through syntax
Kris Cao | Stephen Clark
Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers)

Generating from Abstract Meaning Representation (AMR) is an underspecified problem, as many syntactic decisions are not specified by the semantic graph. To explicitly account for this variation, we break down generating from AMR into two steps: first generate a syntactic structure, and then generate the surface form. We show that decomposing the generation process this way leads to state-of-the-art single model performance generating from AMR without additional unlabelled data. We also demonstrate that we can generate meaning-preserving syntactic paraphrases of the same AMR graph, as judged by humans.


pdf bib
Proceedings of the Third Workshop on Representation Learning for NLP
Isabelle Augenstein | Kris Cao | He He | Felix Hill | Spandana Gella | Jamie Kiros | Hongyuan Mei | Dipendra Misra
Proceedings of the Third Workshop on Representation Learning for NLP


Latent Variable Dialogue Models and their Diversity
Kris Cao | Stephen Clark
Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers

We present a dialogue generation model that directly captures the variability in possible responses to a given input, which reduces the ‘boring output’ issue of deterministic dialogue models. Experiments show that our model generates more diverse outputs than baseline models, and also generates more consistently acceptable output than sampling from a deterministic encoder-decoder model.


A Joint Model for Word Embedding and Word Morphology
Kris Cao | Marek Rei
Proceedings of the 1st Workshop on Representation Learning for NLP