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.CLPaper16d ago

PLC-DPO: Posterior Label Correction in Noisy and Ambiguous Preference Optimization

This matters if you're doing RLHF or DPO at scale and dealing with imperfect human feedback or weak signals. The routing approach (clean/flip/tie) is a practical improvement over naive filtering, and the 60.5 vs 55.5 win rate delta is material. It's not a paradigm shift, but if you're actively training models on preference pairs, revisiting your label-handling strategy here pays off.

alignment rlhf training dpo

arXiv cs.LGPaper20d ago

Disentangling Optimization Scale from Preference Scale in DPO

Anyone tuning alignment pipelines with DPO has probably fought this exact symptom without naming it. The proposed centered-softplus reformulation could make beta sweeps and learning-rate schedules actually predictable, which is a real practical win for teams doing preference tuning in-house.

research llm alignment dpo