OmniTrace converts token-level signals into span-level cross-modal attributions for open-ended generation in omni-modal LLMs via generation-time tracing.
hub
Quantifying Attention Flow in Transformers
15 Pith papers cite this work, alongside 21 external citations. Polarity classification is still indexing.
15
Pith papers citing it
21
external citations · OpenAlex
hub tools
citation-role summary
background 1
method 1
citation-polarity summary
years
2026 15representative citing papers
Debiasing language-model training data for a target group frequently increases stereotyping or counter-stereotyping for non-target groups across categories, models, and scales.
TimeProVe proposes a propose-then-verify framework using lightweight action-based candidate evidence generation followed by targeted VLM verification for efficient long video temporal reasoning, achieving 7.3% improvement on OTB with 75% fewer VLM calls.
Unpack decomposes transformer credit via a unified backward recursion on the φ(S)U template, recovering known IOI circuits with mode labels and showing consistent duplicate-name suppression across Pythia scales from a single forward pass.
Existing visual attribution methods often fail to identify the visual evidence used by LVLMs in chest X-ray reasoning, while MedFocus using unbalanced optimal transport and targeted interventions substantially outperforms them across multiple models and settings.
Low-resolution data improves high-resolution model performance when high-resolution samples are limited, via KL-divergence bounds and experiments on vision transformers and CNNs.
Backward attribution is reframed as integrals over trajectories in a two-player game on the network, unifying gradients and alpha-beta-LRP while enabling new adaptations that outperform prior methods on ViT-B/16 localization metrics.
Personalized soft prompts steer VLM attention to match user-specific gaze patterns, yielding better attention alignment and click prediction in recommendation simulations.
CAAP produces patch attributions in ViTs by direct activation patching on intermediate layers to measure causal contribution to the target class score.
Empirical Jacobian analysis reveals that token influence in trained language models decays as a power law with distance (exponent ~0.8), a learned property not present in random models.
G-Long uses graph-enhanced triplet memory and attention-aware scoring from a T5 summarizer to achieve up to 9.8% better response quality on MSC and 40.8% better retrieval recall on LME with lower overhead.
Training a mean-field Transformer under L2 regularization induces an escape from attention-driven token clustering in later layers after initial clustering.
SAILRec uses dual-side semantic alignment and hierarchical attention steering to improve how LLMs incorporate collaborative embeddings for recommendations, outperforming baselines on MovieLens-1M and Amazon-Book datasets.
Grad-ECLIP is an equivalent but flawed variant of attention-based interpretation, with two principles proposed to ensure model explanations reflect the original model.
Prospective single-center validation of a cascade deep learning dermoscopy CDSS found no false negatives for five malignant lesions and 88.3% specificity, with quantitative IoU assessment of attention maps.
citing papers explorer
-
Through Their Eyes: Fixation-aligned Tuning for Personalized User Emulation
Personalized soft prompts steer VLM attention to match user-specific gaze patterns, yielding better attention alignment and click prediction in recommendation simulations.