Many evaluations of model outputs rely either on contracts checkable at evaluation time or on feedback that arrives within the operating loop. We study the complementary setting in which ground truth is delayed, censored, or private, so deterministic code cannot check correctness at scoring time and must instead issue a code-owned provisional forecast.
arXiv:2607. 10972v1 Announce Type: new Abstract: Many evaluations of model outputs rely either on contracts checkable at evaluation time or on feedback that arrives within the operating loop.
arXiv:2608. 14903v1 Announce Type: new Abstract: Quantitative forecasts of frontier artificial intelligence often connect dated targets to trends in benchmark scores, training compute, release time, or expert belief.
The paper investigates when forecasting agents should employ different behaviors—retrieval, reasoning, deferring to market priors, or using historical analogs—on binary forecasting tasks. It finds that the optimal mechanism depends on the data source, with structured analogs excelling for some processes and market or conservative baselines for others. The authors propose ReliabilityRoute, a rule‑based system that steers agent behavior using reliability features, achieving competitive performance across multiple LLM versions while highlighting that more reasoning is not always better.
The paper introduces loss‑conditioned state execution, a model‑agnostic technique that decides whether to apply a world model’s proposed state change or keep the current state based on whether the change reduces downstream loss. It formalizes state movability as the existence of a loss‑reducing feasible correction and constructs loss‑specific proposals from predictive distributions, executing them only when a groupwise lower confidence bound on loss improvement is positive. Experiments on forecasting and dynamics benchmarks show that the method accepts updates for a subset of cases, achieving lower bounded loss than persistence or always executing the proposal, and highlights that event predictability and loss‑based decisions must be evaluated separately.
arXiv:2607. 12986v1 Announce Type: new Abstract: Plan evaluators can reward a strategic plan for becoming less explicit.
The paper demonstrates that the outcome of a forecasting leaderboard is largely determined by the evaluator’s design choices rather than the models themselves. By fixing the data, horizon, and period, the authors varied three key evaluation decisions—unit of analysis, error pooling, and scoring metric—and showed that each can reverse or eliminate the apparent superiority of any forecasting method. The study also evaluates the practical impact of these choices on a deployed system, revealing that the selection rule captures a significant portion of the potential performance gain, and confirms the findings on an external public dataset.
The paper examines safety routers—systems that route user requests to different language models—and finds that their performance degrades significantly when evaluated under distribution shift. In standard benchmarks, routers appear effective because the best single model is chosen from the same evaluation data, but when the data distribution changes, the routing advantage diminishes or disappears. The study quantifies this bias across multiple safety corpora, showing that routers offer little benefit under realistic shift conditions and that recognition‑based defenses can be undermined by attackers who know the model being used.
The paper demonstrates that emergent capabilities in machine learning models can be forecasted with lead time, calibrated uncertainty, and controlled false‑alarm rates. Using per‑seed analysis on transformers, the authors show that the formation time of a previous‑token head predicts the emergence of an induction head with Spearman ρ = 0.977 and a median lead of 975 training steps. Conformal intervals, blind pre‑registered tests, and a multiplicative rule relating anchor and event times further validate the predictive framework across multiple model families and configurations.
arXiv:2607. 25152v1 Announce Type: new Abstract: Long-running autonomous agents plan, act, and judge their own completion without human intervention.
The paper "trajectory-judge: What Outcome-Only LLM Judges Miss on Agent Trajectories" examines the limitations of outcome-only evaluation for large language model agents. Using a deterministic tool‑using support‑desk environment with a scripted oracle policy and a fault injector, the authors compare five different judging approaches—programmatic rules, outcome‑only, step‑rubric at two model sizes, and a self‑consistency ensemble—on metrics such as detection, step localisation, fault typing, calibration, and cost across 400 trajectories. The study finds that outcome‑only judges miss many silent faults and generate false positives, while step‑rubric judges achieve higher recall with no false alarms but at greater cost, and that none of the judges read the final reply, allowing fabricated promises to evade detection. "whyItMatters":"The findings highlight that current production‑default outcome‑only evaluations can overlook critical failures in agent behavior, underscoring the need for more nuanced, step‑level judging methods to ensure reliable LLM agent performance."
arXiv:2609.10494v1 Announce Type: new Abstract: Enterprises deploy systems, not checkpoints. Usable capability depends jointly on weights, serving route, precision, output contract, and harness, yet...
arXiv:2606. 24996v1 Announce Type: new Abstract: Forecasting leaderboards rank models by predictive quality, but their winners are often read as deployment-ready top-1 advice.