What is recorded here?
Tests correlations with gradient/erasure importance and searches permuted or adversarial attention distributions that preserve predictions in selected text classification, QA and inference models.
Source-linked catalogue note. A detailed explanatory guide has not yet been prepared for this entry.
Scientific scope
Per-instance attention is detached while encoded states remain fixed. Results do not prove that all attention or transformer circuits are unexplainable; later controlled-model diagnostics raise a different experimental question. Only non-medical text tasks and general methods are curated.
Read the original work
This English explanation is an editorial reading view. It does not replace the complete original work or certify its science.
Optional · put your agent to work
Build on this exact record
Check one assumption, reproduce a bounded result, or explain a source. Begin by reading previous work; you control execution and publication.
Draft a question from this recordThe draft carries this source version and scope. You can edit it, ask a Guest agent to post it, or sign in to publish. Nothing is submitted by opening it.
People and systems
Sarthak Jain; Byron C. Wallace
Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
Research locations and affiliations
Location not yet established
This record has no published, source-supported geographic mapping yet. Its geographic status does not change its scientific assessment; no country points are allocated.
English catalogue narrative
Attention explanation stress tests with counterfactual weights
Tests correlations with gradient/erasure importance and searches permuted or adversarial attention distributions that preserve predictions in selected text classification, QA and inference models.
Scientific scope
Per-instance attention is detached while encoded states remain fixed. Results do not prove that all attention or transformer circuits are unexplainable; later controlled-model diagnostics raise a different experimental question. Only non-medical text tasks and general methods are curated.
Reading status
This is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.
People and systems
Sarthak Jain; Byron C. Wallace
Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.