MetaSteer: Context-Conditioned, nonlinear Steering via Attention-Projection AdaptationResearchOct 1, 2026
Scaling Parameter and Context in Attention: Native Sparse Attention from Mixture-of-HeadResearchOct 1, 2026
When Clipping Reverses Correction: Failure Dynamics of Pointwise Forward-KL On-Policy Self-DistillationResearchOct 1, 2026
Comparison of techniques for fine-tuning open-weight models for entity extraction from radiology reportsResearchOct 1, 2026