Subliminal learning occurs via compatible auxiliary and class output heads on task-unrelated inputs, even with random hidden layers or architecture changes, with theory and upper bounds on failure.
Shared global and local geometry of language model embeddings
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
years
2026 4verdicts
UNVERDICTED 4representative citing papers
Unimodal model representations converge to a relational structure captured by the Indra representation via V-enriched Yoneda embedding, which is unique and structure-preserving and improves cross-model and cross-modal robustness when instantiated with angular distance.
RDP-selected 13 layers for LoRA on Qwen3-8B-Base reach 81.67% on MMLU-Math, beating full 36-layer adaptation at 79.32% and random 13-layer selection at 75.56%.
Centroid erasure shows language representations overshadow vision in multimodal models, and text-centroid contrastive decoding recovers substantial accuracy on visual reasoning tasks.
citing papers explorer
-
Learning Through Noise: Why Subliminal Learning Works and When It Fails
Subliminal learning occurs via compatible auxiliary and class output heads on task-unrelated inputs, even with random hidden layers or architecture changes, with theory and upper bounds on failure.
-
The Indra Representation Hypothesis for Multimodal Alignment
Unimodal model representations converge to a relational structure captured by the Indra representation via V-enriched Yoneda embedding, which is unique and structure-preserving and improves cross-model and cross-modal robustness when instantiated with angular distance.
-
RDP LoRA: Geometry-Driven Identification for Parameter-Efficient Adaptation in Large Language Models
RDP-selected 13 layers for LoRA on Qwen3-8B-Base reach 81.67% on MMLU-Math, beating full 36-layer adaptation at 79.32% and random 13-layer selection at 75.56%.
-
The Cost of Language: Centroid Erasure Exposes and Exploits Modal Competition in Multimodal Language Models
Centroid erasure shows language representations overshadow vision in multimodal models, and text-centroid contrastive decoding recovers substantial accuracy on visual reasoning tasks.