When Only Time Will Tell: Interpreting How Transformers Process Local Ambiguities Through the Lens of Restart-Incrementality
This work addresses the challenge of interpreting model revisions in natural language processing for researchers, but it is incremental as it builds on existing restart-incremental methods.
The paper tackled the problem of how restart-incremental Transformers process local ambiguities in sentences, showing that their sequential states encode information on the garden path effect and its resolution, contributing to their advantage over causal models in handling revisions.
Incremental models that process sentences one token at a time will sometimes encounter points where more than one interpretation is possible. Causal models are forced to output one interpretation and continue, whereas models that can revise may edit their previous output as the ambiguity is resolved. In this work, we look at how restart-incremental Transformers build and update internal states, in an effort to shed light on what processes cause revisions not viable in autoregressive models. We propose an interpretable way to analyse the incremental states, showing that their sequential structure encodes information on the garden path effect and its resolution. Our method brings insights on various bidirectional encoders for contextualised meaning representation and dependency parsing, contributing to show their advantage over causal models when it comes to revisions.