Dimension d = O(m^{-2} log n) nearly achieves the optimal margin m^rd(+∞, A) for retrieval embeddings, with matching lower bounds showing d = O(k log(n/k)) suffices and is necessary for m = Θ(k^{-1/2}) on k-sparse query matrices.
Title resolution pending
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
representative citing papers
Dual-view training via polarity reversal improves instruction-following retrieval performance by 45% on the FollowIR benchmark using a 305M-parameter encoder.
Projection heads act as geometric buffers; nonlinear heads induce negative Hessian curvature to escape dimensional collapse while linear heads rely on discrete dynamics and BatchNorm.
LLMs contain identifiable COCO neurons that enable implicit self-correction against stereotypes; targeted editing of these neurons improves fairness and robustness to jailbreaks while preserving generation quality.
Context-Aligned Contrastive Regression combines cross-view context alignment and ordinal soft contrastive learning with ridge ensembles to improve lexical difficulty prediction across L1 backgrounds on three datasets.
Nomic AI produced and open-sourced a reproducible 8192-context English text embedder that exceeds OpenAI Ada-002 and text-embedding-3-small performance on MTEB short-context and LoCo long-context benchmarks.
LLMs exhibit compartmentalization by learning separate internal representations for equivalent concepts presented differently, which reduces sample efficiency and resists unification even with synthetic parallel data.
citing papers explorer
-
Is Dimensionality a Barrier for Retrieval Models?
Dimension d = O(m^{-2} log n) nearly achieves the optimal margin m^rd(+∞, A) for retrieval embeddings, with matching lower bounds showing d = O(k log(n/k)) suffices and is necessary for m = Θ(k^{-1/2}) on k-sparse query matrices.
-
Dual-View Training for Instruction-Following Information Retrieval
Dual-view training via polarity reversal improves instruction-following retrieval performance by 45% on the FollowIR benchmark using a 305M-parameter encoder.
-
The Geometry of Projection Heads: Conditioning, Invariance, and Collapse
Projection heads act as geometric buffers; nonlinear heads induce negative Hessian curvature to escape dimensional collapse while linear heads rely on discrete dynamics and BatchNorm.
-
Modeling Implicit Conflict Monitoring Mechanisms against Stereotypes in LLMs
LLMs contain identifiable COCO neurons that enable implicit self-correction against stereotypes; targeted editing of these neurons improves fairness and robustness to jailbreaks while preserving generation quality.
-
Improving Lexical Difficulty Prediction with Context-Aligned Contrastive Learning and Ridge Ensembling
Context-Aligned Contrastive Regression combines cross-view context alignment and ordinal soft contrastive learning with ridge ensembles to improve lexical difficulty prediction across L1 backgrounds on three datasets.
-
Nomic Embed: Training a Reproducible Long Context Text Embedder
Nomic AI produced and open-sourced a reproducible 8192-context English text embedder that exceeds OpenAI Ada-002 and text-embedding-3-small performance on MTEB short-context and LoCo long-context benchmarks.
-
Language models struggle with compartmentalization
LLMs exhibit compartmentalization by learning separate internal representations for equivalent concepts presented differently, which reduces sample efficiency and resists unification even with synthetic parallel data.
- Where Does Authorship Signal Emerge in Encoder-Based Language Models?