Margin Dynamics Explain DPO Divergence and a Finite SimPO Attractor
Margin Dynamics for Large Language Model Alignment frames pairwise preference optimization…
ZICQ curates daily AI news and deep dives on models, open-source infrastructure, and developer field reports.
Submit an articleCurrent filter: Tag: #AdaFlowLite Clear filters
Margin Dynamics for Large Language Model Alignment frames pairwise preference optimization…
Research & Papers· 2026-11-18
The paper turns a layerwise learning-rate profile into a single, computable effective rate…
Research & Papers· 2026-11-18
AdaFlowLite replaces barrier-style multimodal fusion with opportunistic inference: process…
Research & Papers· 2026-11-18
NEBench R1 converts U.S. NRC reactor-operator licensing examinations into a 7,546-question…
Research & Papers· 2026-11-18
论文发布危机咨询评测套件CrisisBench,以200段去标识化热线对话构造1,085个单轮任务,并用热线元数据生成1,185个患者LLM,配合经专家校准的23项LLM裁判。五个…
Agentic· 2026-11-18
DeltaHarness treats the agent execution harness as an adaptive control layer rather than a…
Research & Papers· 2026-11-17
Heave-Ho presents a decentralized architecture for long-horizon loco-manipulation across m…
Research & Papers· 2026-11-17
TCP4D addresses robot manipulation learning from monocular RGB videos by representing acti…
Research & Papers· 2026-11-17
Pi4 unifies single images, unordered image collections, and videos under a roughly 0.5B-pa…
Research & Papers· 2026-11-17
The paper reframes model construction as weight-space programming: instead of optimizing p…
Industry & Trends· 2026-11-17
Nvidia’s interim CEO has reportedly reaffirmed the company’s open AI strategy following Je…
Compute & Architecture· 2026-11-17
Amazon SageMaker HyperPod Inference now supports managed NVIDIA Dynamo, bringing multi-nod…