Research radar

arXiv preprints, filtered for developer impact.

Concise summaries of new AI papers, ranked for how likely they are to change how production systems are built. arXiv collection remains rate-limited.

Papers tracked

3,367

matching records

Shown

current page

Top radar

http://arxiv.org/abs/2610.02173v1

Code links

on this page

Paper	Authors	arXiv ID	Categories	Published	Code	Radar	Summary
Every Ablation Is a Dose: Counterweights and the Semblance of Self-Repair	Areeb Ahmad, Pratinav Seth, Vinay Kumar Sankarapu	http://arxiv.org/abs/2610.02173v1	cs.LG, cs.CL	Oct 1, 2026	None	Score61	Ablate a component of a language model, and other components often appear to adjust and compe...
Disentangling Computation in Multi-Task Neural Networks with the Green's Operator	James Hazelden	http://arxiv.org/abs/2609.40292v1	cs.LG, q-bio.NC	Sep 30, 2026	None	Score63	How is computation organized and reused across tasks and time in a trained recurrent network?...
FurE: Efficient Instance-Specific 3D Fur Reconstruction without Animal-Fur Datasets	Srinjay Sarkar, Prakhar Kaushik, Soumava Paul	http://arxiv.org/abs/2609.35770v1	cs.CV, cs.AI	Sep 28, 2026	None	Score81	Realistic and editable animal fur reconstruction from multi-view images is challenging due to...
Telescopic Language Models	Zhilin Guo, Boqiao Zhang, Hakan Aktas	http://arxiv.org/abs/2609.35769v1	cs.CL, cs.AI	Sep 28, 2026	None	Score66	One deployed language model must often serve many compute budgets, yet serving each budget st...
KV-streams for Efficient Compaction in Agentic Reinforcement Learning	Emiliano Penaloza, Dane Malenfant, Dheeraj Vattikonda	http://arxiv.org/abs/2609.35750v1	cs.LG, cs.AI	Sep 28, 2026	None	Score65	Scaling the horizon of agentic LLMs is bottlenecked by the need to fit ever longer context tr...
Copy the Same, Distill the Difference: Initializing Linear Vision Transformers	Huaiyuan Qin, Muli Yang, Gabriel James Goenawan	http://arxiv.org/abs/2609.35745v1	cs.CV, cs.AI	Sep 28, 2026	None	Score76	Linear Vision Transformers (ViTs) are designed to replace the attention in Softmax ViTs with...
When May a Bandit Leave Its Anchor? E-Process-Authorized Thompson Sampling under Non-stationarity	Mayand Gulati, Kerong Wang, WeiChen Au	http://arxiv.org/abs/2610.03646v1	cs.LG, stat.ML	Oct 2, 2026	None	Score61	Stationarity rewards memory, but after a change the same history can mislead. We ask when for...
On the Convergence of Success Conditioning for Policy Optimization	Matthew Brun, Xu Andy Sun	http://arxiv.org/abs/2610.03642v1	cs.LG, math.OC	Oct 2, 2026	None	Score61	Success conditioning is a strategy for improving decision-making policies in stochastic envir...
Statistical Learning of Contractive Dynamical Representations for Composite Adaptive Control	Min Kim, José Leonardo Brenes, Fred Hadaegh	http://arxiv.org/abs/2609.35758v1	eess.SY, cs.LG	Sep 28, 2026	None	Score74	We present a representation-learning framework for composite adaptive tracking control under...
Private online learning and prediction for Littlestone classes	Amartya Sanyal	http://arxiv.org/abs/2610.06822v1	cs.LG, cs.CR	Oct 5, 2026	None	Score64	We study mistake bounds for differentially private online learning and online prediction unde...