KPKoyena Pal
Koyena Pal
*This is an algorithmically generated author record based on publicly available information.
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Koyena Pal has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Andrew Yuan
Northeastern University Northeastern University UMass Amherst Northeastern University Northeastern University
2 shared publications - Byron C. Wallace
Northeastern University
2 shared publications - David Bau
Northeastern University
2 shared publications - Jiuding Sun2 shared publications
Latest Publications
- Open Access
Future Lens: Anticipating Subsequent Tokens from a Single Hidden State
We conjecture that hidden state vectors corresponding to individual input tokens encode information sufficient to accurately predict several tokens ahead. More concretely, in this paper we ask: Given a hidden (internal) representation of a single token at position $t$ in an input, can we reliably anticipate the tokens that will appear at positions $\geq t + 2$? To test this, we measure linear approximation and causal intervention methods in GPT-J-6B to evaluate the degree to which individual hidden states in the network contain signal rich enough to predict future hidden states and, ultimately, token outputs. We find that, at some layers, we can approximate a model's output with more than 48% accuracy with respect to its prediction of subsequent tokens through a single hidden state. Finally we present a "Future Lens" visualization that uses these methods to create a new view of transformer states.Open AccessFuture Lens: Anticipating Subsequent Tokens from a Single Hidden State
We conjecture that hidden state vectors corresponding to individual input tokens encode information sufficient to accurately predict several tokens ahead. More concretely, in this paper we ask: Given a hidden (internal) representation of a single token at position $t$ in an input, can we reliably anticipate the tokens that will appear at positions $\geq t + 2$? To test this, we measure linear approximation and causal intervention methods in GPT-J-6B to evaluate the degree to which individual hidden states in the network contain signal rich enough to predict future hidden states and, ultimately, token outputs. We find that, at some layers, we can approximate a model's output with more than 48% accuracy with respect to its prediction of subsequent tokens through a single hidden state. Finally we present a "Future Lens" visualization that uses these methods to create a new view of transformer states. - Top Cited PapersOpen Access
- 1 January 2023
- Published by Association for Computational Linguistics (ACL)
- p. 548-560
Future Lens: Anticipating Subsequent Tokens from a Single Hidden State
Koyena Pal, Jiuding Sun, Andrew Yuan, Byron Wallace, David Bau. Proceedings of the 27th Conference on Computational Natural Language Learning (CoNLL). 2023.
- 1 January 2023
- Published by Association for Computational Linguistics (ACL)
- p. 548-560
Top Cited PapersOpen AccessFuture Lens: Anticipating Subsequent Tokens from a Single Hidden State
Koyena Pal, Jiuding Sun, Andrew Yuan, Byron Wallace, David Bau. Proceedings of the 27th Conference on Computational Natural Language Learning (CoNLL). 2023.