# When Only Time Will Tell: Interpreting How Transformers Process Local Ambiguities Through the Lens of Restart-Incrementality

arXiv: https://arxiv.org/abs/2402.13113 · HTML: https://mlsanity.com/paper/2402.13113/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 44 of its 90 references that are in the map.
- First posted on arXiv 20 Feb 2024.

## Builds on (most cited first, 30 of 44)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [RoBERTa: A Robustly Optimized BERT Pretraining Approach](https://mlsanity.com/paper/1907.11692/) (2019)
- [OPT: Open Pre-trained Transformer Language Models](https://mlsanity.com/paper/2205.01068/) (2022)
- [BERT Rediscovers the Classical NLP Pipeline](https://mlsanity.com/paper/1905.05950/) (2019)
- [What Does BERT Look At? An Analysis of BERT's Attention](https://mlsanity.com/paper/1906.04341/) (2019)
- [A Primer in BERTology: What we know about how BERT works](https://mlsanity.com/paper/2002.12327/) (2020)
- [Quantifying Attention Flow in Transformers](https://mlsanity.com/paper/2005.00928/) (2020)
- [Stanza: A Python Natural Language Processing Toolkit for Many Human Languages](https://mlsanity.com/paper/2003.07082/) (2020)
- [Eliciting Latent Predictions from Transformers with the Tuned Lens](https://mlsanity.com/paper/2303.08112/) (2023)
- [Transformer Feed-Forward Layers Build Predictions by Promoting Concepts in the Vocabulary Space](https://mlsanity.com/paper/2203.14680/) (2022)
- [Linguistic Knowledge and Transferability of Contextual Representations](https://mlsanity.com/paper/1903.08855/) (2019)
- [Deep Biaffine Attention for Neural Dependency Parsing](https://mlsanity.com/paper/1611.01734/) (2016)
- [Analysis Methods in Neural Language Processing: A Survey](https://mlsanity.com/paper/1812.08951/) (2018)
- [Neural Language Models as Psycholinguistic Subjects: Representations of Syntactic State](https://mlsanity.com/paper/1903.03260/) (2019)
- [We're Afraid Language Models Aren't Modeling Ambiguity](https://mlsanity.com/paper/2304.14399/) (2023)
- [Sudden Drops in the Loss: Syntax Acquisition, Phase Transitions, and Simplicity Bias in MLMs](https://mlsanity.com/paper/2309.07311/) (2023)
- [Future Lens: Anticipating Subsequent Tokens from a Single Hidden State](https://mlsanity.com/paper/2311.04897/) (2023)
- [Jump to Conclusions: Short-Cutting Transformers With Linear Transformations](https://mlsanity.com/paper/2303.09435/) (2023)
- [Back to the Future: Unsupervised Backprop-based Decoding for Counterfactual and Abductive Commonsense Reasoning](https://mlsanity.com/paper/2010.05906/) (2020)
- [Machine Translation Decoding beyond Beam Search](https://mlsanity.com/paper/2104.05336/) (2021)
- [Problems with Cosine as a Measure of Embedding Similarity for High Frequency Words](https://mlsanity.com/paper/2205.05092/) (2022)
- [Explaining How Transformers Use Context to Build Predictions](https://mlsanity.com/paper/2305.12535/) (2023)
- [Viable Dependency Parsing as Sequence Labeling](https://mlsanity.com/paper/1902.10505/) (2019)
- [A Targeted Assessment of Incremental Processing in Neural LanguageModels and Humans](https://mlsanity.com/paper/2106.03232/) (2021)
- [Re-translation versus Streaming for Simultaneous Translation](https://mlsanity.com/paper/2004.03643/) (2020)
- [Syntactic Surprisal From Neural Models Predicts, But Underestimates, Human Processing Difficulty From Syntactic Ambiguities](https://mlsanity.com/paper/2210.12187/) (2022)
- [Re-Translation Strategies For Long Form, Simultaneous, Spoken Language Translation](https://mlsanity.com/paper/1912.03393/) (2019)
- [Probing for Incremental Parse States in Autoregressive Language Models](https://mlsanity.com/paper/2211.09748/) (2022)
- [Looking Right is Sometimes Right: Investigating the Capabilities of Decoder-only LLMs for Sequence Labeling](https://mlsanity.com/paper/2401.14556/) (2024)

## Built on by (most cited first, 1 of 1)

- [Prior Lessons of Incremental Dialogue and Robot Action Management for the Age of Language Models](https://mlsanity.com/paper/2501.00953/) (2025)

## Abstract (arXiv)

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.