Agentic workloads are a major source of LLM inference demand. Existing prefill--decode (PD) disaggregated serving systems offer limited flexibility for runtime adaptation of compute allocation and KV cache management, and rely on coarse signals that only partly capture the diverse and dynamic agent workload. These limi...
Hao Kang, Wei-Li Xu, Zi-Yang Li et al.· 0 citations
Results show that how information is represented to an agent shapes both its cost and its success, making the environment worth optimizing in addition to the parameters and the harness, making the environment worth optimizing in addition to the parameters and the harness.
Ming-Xuan Li, Huan-Zhi Mao, Qiu-Yang Mang et al.· 0 citations
TwinRouterBench is introduced, a step-level routing benchmark with two tracks that supports fast offline iteration followed by end-to-end validation under live agent execution, and success is measured by official task resolution and realized API spend.
Pei Yang, Wan-Yi Chen, Tong Yang et al.· arXiv.org· 6 citations
Combee is proposed, a novel framework to scale parallel prompt learning for self-improving agents that speeds up learning and enables running many agents in parallel while learning from their aggregate traces without quality degradation.
This work defines the scaling inflection point as the per-session budget where marginal Elo gains match the independent-sampling reference, and proposes Elo-per-token analysis, which tracks the best solution found at each token budget and uses a Bradley-Terry model to aggregate within-task orderings into Elo ratings ac...