Tootfinder

Beyond Turn Limits: Training Deep Search Agents with Dynamic Context Window
Qiaoyu Tang, Hao Xiang, Le Yu, Bowen Yu, Yaojie Lu, Xianpei Han, Le Sun, WenJuan Zhang, Pengbo Wang, Shixuan Liu, Zhenru Zhang, Jianhong Tu, Hongyu Lin, Junyang Lin
https://arxiv.org/abs/2510.08276

Beyond Turn Limits: Training Deep Search Agents with Dynamic Context Window
While recent advances in reasoning models have demonstrated cognitive behaviors through reinforcement learning, existing approaches struggle to invoke deep reasoning capabilities in multi-turn agents with long-horizon interactions. We propose DeepMiner, a novel framework that elicits such abilities by introducing high-difficulty training tasks and dynamic context window. DeepMiner presents a reverse construction method to generate complex but verifiable question-answer pairs from authentic web …

Don't miss today's packed Metacurity for the most critical infosec developments you should know, including
--Thai police blew up a Myanmar cyberscam compound,
--FCC blocks more Chinese-made devices,
--Hacking Team successor surfaces,
--Aisuru botnet overhauled to rent out IoT devices,
--Herodotus Android trojan mimics human behavior,
--Ad giant's subsidiary exposed data in cyber incident,
--CBP searches more devices than ever,
--ICE …

Thai police blew up a Myanmar cyberscam compound
FCC blocks more Chinese-made devices, Hacking Team successor surfaces, Aisuru botnet overhauled to rent out IoT devices, Herodotus Android trojan mimics human behavior, Ad giant's subsidiary exposed data in cyber incident, CBP searches more devices than ever, ICE is becoming a spymaster, much more

Came across this 2010 blog post about mindfulness in computing and so much of these behaviors have only intensified to new extremes with LLM usage. So much so that not only is the process of software creation being quickly supplanted by prompts and (stochastic) "search" assemblies, but more generally the kind of mindfulness talked about in the post (here meaning thinking through & solving a problem yourself[1]) is now being openly discouraged by industry and forcefully delegate…

Representation-Based Exploration for Language Models: From Test-Time to Post-Training
Jens Tuyls, Dylan J. Foster, Akshay Krishnamurthy, Jordan T. Ash
https://arxiv.org/abs/2510.11686

Representation-Based Exploration for Language Models: From Test-Time to Post-Training
Reinforcement learning (RL) promises to expand the capabilities of language models, but it is unclear if current RL techniques promote the discovery of novel behaviors, or simply sharpen those already present in the base model. In this paper, we investigate the value of deliberate exploration -- explicitly incentivizing the model to discover novel and diverse behaviors -- and aim to understand how the knowledge in pre-trained models can guide this search. Our main finding is that exploration wi…