Benjamin Van Durme

Seq vs Seq: An Open Suite of Paired Encoders and Decoders

SEQR: Secure and Efficient QR-based LoRA Routing

mmBERT: A Modern Multilingual Encoder with Annealed Language Learning

The Alignment Waltz: Jointly Training Agents to Collaborate for Safety

When Knowledge Hurts: Enriching Domain Knowledge for Causal Scientific Reasoning

Conformal Linguistic Calibration: Trading-off between Factuality and Specificity

Multi-Field Adaptive Retrieval

Always Tell Me The Odds: Fine-grained Conditional Probability Estimation

Controllable Safety Alignment: Inference-Time Adaptation to Diverse Safety Requirements

Rank1: Test-Time Compute for Reranking in Information Retrieval

From Models to Microtheories: Distilling a Model's Topical Knowledge for Grounded Question-Answering

Promptriever: Instruction-Trained Retrievers Can Be Prompted Like Language Models

SpectR: Dynamically Composing LM Experts with Spectral Routing

Generative Adapter: Contextualizing Language Models in Parameters with A Single Forward Pass

Hierarchical Corpus Encoder: Fusing Generative Retrieval and Dense Indices

RE-Adapt: Reverse Engineered Adaptation of Large Language Models

KV-Distill: Nearly Lossless Context Compression for Transformers