Topic
post-training
AAPA: Adversarially Anchored Preference Alignment Enhances LLM Post-Training Performance
Researchers propose AAPA, a plug-in framework that adds a sentence-level adversarial anchoring signal to existing post-training objectives for large language models. Experiments on instruction-following benchmarks show consistent improvements, with staged AAPA achieving 5.77% gain on Qwen3-0.6B and 3.75% on Qwen3-4B over a strong GRPO baseline.
Which Pairs to Compare for LLM Post-Training? Research Reveals Optimal Labeling Strategy
A new arXiv paper by researchers Han, Goyal, and Ma addresses the challenge of which comparison pairs to label in preference-based LLM post-training. The study formulates comparison curation as a sampling-design problem and provides theoretical bounds showing how selection affects policy performance. Experiments demonstrate that proposed designs improve sample efficiency over common heuristics.
ReGrad: A New AI Paradigm for Continual Learning Without Catastrophic Forgetting
A new paper introduces ReGrad (Retrievable Gradients), a paradigm for continual post-training that pre-computes document-specific gradients, stores them in a Gradient Bank, and retrieves query-relevant gradients at inference time for temporary weight adaptation. The method uses bi-level meta-learning to reshape gradients into generalizable signals, outperforming CPT and RAG baselines in experiments.