AI Paper Atlas

Research map / Large language models / Reasoning and chain of thought

AiR: Attention with Reasoning Capability

Shi Chen, Ming Jiang, Jinhui Yang, Qi Zhao

ECCV 2020 · Large language models › Reasoning and chain of thought · approach: Long CoT dynamics and efficiency

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add AiR-M progressive attention supervisionUpDown · GQA test-devaccuracy (%): 51.31 -> 53.46Improved
add AiR-M progressive attention supervisionMUTAN · GQA test-devaccuracy (%): 50.78 -> 51.81Improved
add AiR-M progressive attention supervisionBAN · GQA test-devaccuracy (%): 50.14 -> 53.36Improved
add PAAN adversarial attention supervisionUpDown · GQA test-devaccuracy (%): 51.31 -> 48.03Hurt
add HAN human-like attention supervisionUpDown · GQA test-devaccuracy (%): 51.31 -> 49.96Hurt
add ASM supervision from objects mined from languageUpDown · GQA test-devaccuracy (%): 51.31 -> 52.96Improved
use progressive multi-step supervision instead of single-glimpse joint supervision (ablation)AiR-M on UpDown · GQA test-devaccuracy (%): 52.84 -> 53.46Improved
add attention supervision L_alpha to progressive reasoning-operation supervision (ablation)AiR-M on UpDown · GQA test-devaccuracy (%): 50.01 -> 53.46Improved
add reasoning-operation supervision L_r to progressive attention supervision (ablation)AiR-M on UpDown · GQA test-devaccuracy (%): 50.33 -> 53.46Improved
increase operation loss weight phi from 0.5 to 10 (ablation)AiR-M on UpDown · GQA balanced valaccuracy (%): 62.57 -> 61.39Hurt
vary phi from 0.01 to 1 (ablation)AiR-M on UpDown · GQA balanced valaccuracy (%): 62.02 / 62.11 / 62.57 / 61.93No clear effect
use object-based attention instead of spatial attention (ablation)UpDown VQA attention · AiR-E on GQA subsetAiR-E (query): 0.725 -> 3.241Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy