9 citations · 17 across the 10 of their papers we have counts for
Showing 2023Show all
2 papers · 1 filter
cs.CL2023★ 9 cited
Future Lens: Anticipating Subsequent Tokens from a Single Hidden State
Koyena Pal, Jiuding Sun, Andrew Yuan +2
We conjecture that hidden state vectors corresponding to individual input tokens encode information sufficient to accurately predict several tokens ahead. More concretely, in this…
cs.CL2023★ 4 cited
Function Vectors in Large Language Models
Eric Todd, Millicent L. Li, Arnab Sen Sharma +3
We report the presence of a simple neural mechanism that represents an input-output function as a vector within autoregressive transformer language models (LMs). Using causal media…