Fine-tuning updates frequently stale activation monitors for language model safety while quantization does not, with degradation predictable and repairable via label-free realignment.
Title resolution pending
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
years
2026 6representative citing papers
Amplifying reasoning task vectors (α>1) surfaces learned secrets in LLMs up to 10× more frequently than standard reasoning models across four secret-keeping settings.
A hybrid first-order then zeroth-order optimization approach improves robustness of safety-aligned LLMs while preserving utility, with layer-wise sensitivity estimation for efficiency.
A single adversary in distributed training inflates its attribution value via latent optimization on synthetic batches without degrading accuracy or triggering basic defenses.
Residual-stream noise injection raises narrative diversity in Arabic educational stories while preserving reading-grade level, outperforming high-temperature sampling across five 7-9B models.
N-GRPO enhances GRPO via Semantic Neighbor Mixing of token embeddings to improve diversity and consistency in LLM math reasoning rollouts.
citing papers explorer
-
Do Activation Monitors Survive Model Updates? Benchmarking, Predicting, and Repairing Activation-Monitor Staleness
Fine-tuning updates frequently stale activation monitors for language model safety while quantization does not, with degradation predictable and repairable via label-free realignment.
-
Overthinking: Amplifying Reasoning Weights to Extract Learned Secrets
Amplifying reasoning task vectors (α>1) surfaces learned secrets in LLMs up to 10× more frequently than standard reasoning models across four secret-keeping settings.
-
Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization
A hybrid first-order then zeroth-order optimization approach improves robustness of safety-aligned LLMs while preserving utility, with layer-wise sensitivity estimation for efficiency.
-
On the Fragility of Data Attribution When Learning Is Distributed
A single adversary in distributed training inflates its attribution value via latent optimization on synthetic batches without degrading accuracy or triggering basic defenses.
-
Noise Steering for Controlled Text Generation: Improving Diversity and Reading-Level Fidelity in Arabic Educational Story Generation
Residual-stream noise injection raises narrative diversity in Arabic educational stories while preserving reading-grade level, outperforming high-temperature sampling across five 7-9B models.
-
N-GRPO: Embedding-Level Neighbor Mixing for Enhanced Policy Optimization
N-GRPO enhances GRPO via Semantic Neighbor Mixing of token embeddings to improve diversity and consistency in LLM math reasoning rollouts.