Large language models (LLMs) appear to exhibit human-level abilities on a range of tasks, yet they are notoriously considered to be “black boxes”, and little is known about the internal representations and mechanisms that underlie their behavior. This talk will discuss recent work which seeks to illuminate the processing that takes place under the hood. I will focus in particular on questions related to LLM’s ability to represent abstract, compositional, and content-independent operations of the type assumed to be necessary for advanced cognitive functioning in humans. 

 

References

Pavlick, Ellie. “Symbols and grounding in large language models.” Philosophical Transactions of the Royal Society A 381.2251 (2023): 20220041. https://royalsocietypublishing.org/doi/pdf/10.1098/rsta.2022.0041

Lepori, Michael A., Thomas Serre, and Ellie Pavlick. “Break it down: evidence for structural compositionality in neural networks.” arXiv preprint arXiv:2301.10884 (2023). https://arxiv.org/pdf/2301.10884.pdf

Merullo, Jack, Carsten Eickhoff, and Ellie Pavlick. “Language Models Implement Simple Word2Vec-style Vector Arithmetic.” arXiv preprint arXiv:2305.16130 (2023). https://arxiv.org/pdf/2305.16130.pdf

Tenney, Ian, Dipanjan Das, and Ellie Pavlick. “BERT Rediscovers the Classical NLP Pipeline.” Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. 2019. https://arxiv.org/pdf/1905.05950.pdf