University of Groningen on Twitter

“Great news for @DocTinaK and @AnaGuerberof! They both got an @ERC_Research #consolidator grant of €2 million for research into, respectively, #parenthood and the creative process of #translating. Congrats!🎀 More 👉 https://t.co/v5o0SBfzu5 @rug_gmw @FacultyofArtsUG”

Quantifying Context Mixing in Transformers

Self-attention weights and their transformed variants have been the main source of information for analyzing token-to-token interactions in Transformer-based models. But despite their ease of interpretation, these weights are not faithful to the models' decisions as they are only one part of an encoder, and other components in the encoder layer can have considerable impact on information mixing in the output representations. In this work, by expanding the scope of analysis to the whole encoder block, we propose Value Zeroing, a novel context mixing score customized for Transformers that provides us with a deeper understanding of how information is mixed at each encoder layer. We demonstrate the superiority of our context mixing score over other analysis methods through a series of complementary evaluations with different viewpoints based on linguistically informed rationales, probing, and faithfulness analysis.

Quantifying Context Mixing in Transformers

Self-attention weights and their transformed variants have been the main source of information for analyzing token-to-token interactions in Transformer-based models. But despite their ease of interpretation, these weights are not faithful to the models' decisions as they are only one part of an encoder, and other components in the encoder layer can have considerable impact on information mixing in the output representations. In this work, by expanding the scope of analysis to the whole encoder block, we propose Value Zeroing, a novel context mixing score customized for Transformers that provides us with a deeper understanding of how information is mixed at each encoder layer. We demonstrate the superiority of our context mixing score over other analysis methods through a series of complementary evaluations with different viewpoints based on linguistically informed rationales, probing, and faithfulness analysis.

Vicki on Twitter

“As Gregor Samsa awoke one morning from uneasy dreams, he found himself rewritten in Rust for performance reasons”

Marco Zullich on Twitter

“Happy to announce that, from February 1st, I'll start a new job as a lecturer 👨‍🏫 at @univgroningen Excited to start this new adventure!”

Tweet / Twitter

Tweet / Twitter

Release v0.3.3: Attention attribution, new aggregation, improved saving/reloading and more · inseq-team/inseq

What’s Changed Attention attribution (#148 ) This release introduces a new category of attention attribution methods and adds support for AttentionAttribution (id: attention). This method attribute...