The Signal

Everything that matters in AI, with our take.

Updated through the day. Every headline links straight to the source. The two lines underneath are ours.

research1156 agents716 infra561 labs317 culture268 capital253 regulation233 claude175 openai149 safety103 security89 evaluation77 reasoning61 alignment52 benchmarks50 interpretability48

arXiv cs.AIPaper8d ago

Understanding the Impact of Model Pruning on Long-Tail Forgetting and Explanation Reliability in Medical Imaging

Model compression is standard practice in medical deployment, but this paper shows the cost is higher than accuracy metrics reveal: rare diseases matter clinically and pruning hits them hardest. If you're shipping a medical model, pruning audits need to stratify by condition frequency and validate explanations separately from predictions. The finding applies broadly anywhere rare-class reliability matters.

research medical-ai model-compression interpretability

arXiv cs.CLPaper14d ago

Scalable Kronecker-Fisher Approximation: Efficient Hessian Analysis for Billion-Parameter Language Models Compression

Compression is about to matter a lot more. This work shows that value projections are consistently fragile across model families, and you can figure out where to cut without brute force. If you're shipping models to edge devices or trying to reduce inference cost, this gives you a principled way to identify where safe sparsity exists. The cross-layer insight is novel enough to change how you think about pruning.

research model-compression quantization efficiency

arXiv cs.CLPaper15d ago

Knowledge Distillation During Mid-Training Favors Reasoning over Factual Recall

This is useful signal if you're training smaller models via distillation. The finding is counterintuitive: the standard KD formulation amplifies teacher confidence disparities across domains, which hurts factual knowledge transfer at the mid-training stage specifically. If you're distilling and notice your model reasons well but has shaky factual grounding, check whether you're running KD during the wrong phase or need domain-specific distillation schemes.

research model-compression distillation

arXiv cs.LGPaper15d ago

A Mathematical Theory of Reusable Neural Bases for Network Compression

This is a compression technique with decent experimental results but no clear advantage over existing methods like LoRA or MoE. The claim of stability is unproven at scale. If you're constrained by memory in training or inference, it's worth a look, but the paper doesn't make a compelling case for why this beats alternatives.

model-compression parameter-efficiency neural-architecture