Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
The pipeline replaces prohibitively expensive manual architectural labeling with a scalable agentic approach, enabling fine-tuned models to achieve dramatically higher SWE-bench Verified resolved rates than either the base model or unfiltered fine-tuning.
LatentGym fills a gap left by existing frameworks by providing the first controllable latent structure and disentangled exploration/exploitation metrics for measuring cross-task experiential learning in LLM agents.
The benchmark demonstrates that tool-menu composition — not just model capability — is a primary driver of agent task success, error rates, and safety-relevant risk exposure, with CMTF cutting token usage by roughly 98% and more than doubling task success over unfiltered baselines.
RSA demonstrates that dynamic, context-targeted auditing catches malicious agent skills that static detectors miss and remain robust under self-evolving adversarial attacks where static methods collapse.
SkillAudit removes the dependency on privileged external feedback signals that existing skill-evolution methods require, enabling agent skill improvement in real-world deployments where only a task description and workspace data are available.
Compilation-based metrics, the current standard for judging autoformalization agents, are shown to substantially overstate quality by missing semantic errors that a reproducible three-dimensional audit framework can detect.
Despite code access giving LLM agents a measurable edge on time series tasks, a 22–34% error rate on benchmark questions exposes a concrete reliability gap that limits their use in high-stakes automated decision-making domains like finance and healthcare.
SING reduces full-corpus tool-schema exposure by 99.8% while simultaneously improving retrieval recall and task success, directly addressing the context-cost and closed-world limitations that arise as agentic tool ecosystems scale to thousands of APIs.
The talk documents a concrete, production-tested eval architecture that closed the loop between offline simulation and live agent behavior at scale, directly enabling Lyft's resolution rate to climb from 10% to 35%.
Fable 5 is the first model to outscore a cherry-picked composite of best-in-class specialists across a full multi-turn SDLC workflow on Ship-Bench, though the nearly $180 API cost the article documents frames its viability as an open cost-versus-reliability question.