Vitoria Guardieiro
@viguardieiro.bsky.social
PhD Student @ UPenn CIS
Reposted by Vitoria Guardieiro
Excited to share our new paper: "Instruction Following by Boosting Attention of Large Language Models"!
We introduce Instruction Attention Boosting (InstABoost), a simple yet powerful method to steer LLM behavior by making them pay more attention to instructions.
(🧵1/7)
We introduce Instruction Attention Boosting (InstABoost), a simple yet powerful method to steer LLM behavior by making them pay more attention to instructions.
(🧵1/7)
July 10, 2025 at 6:22 PM
Excited to share our new paper: "Instruction Following by Boosting Attention of Large Language Models"!
We introduce Instruction Attention Boosting (InstABoost), a simple yet powerful method to steer LLM behavior by making them pay more attention to instructions.
(🧵1/7)
We introduce Instruction Attention Boosting (InstABoost), a simple yet powerful method to steer LLM behavior by making them pay more attention to instructions.
(🧵1/7)