Replacing one LLM agent with a collaborating team can raise accuracy, but whether scaling the team helps, and which architecture to scale, is unclear. Sweeping eight agent orchestration architectures across five instruction-tuned 7-9B models, five short-answer benchmarks, and an executable-code benchmark up to 30 calls...
This work introduces Steiner's taxonomy of group tasks as a framework for analyzing multi-agent LLM scaling and shows that task structure is a fundamental determinant of team scaling.
This work formulate dense univariate segmentation as node classification on a graph whose nodes are the original time points as node classification on a graph whose nodes are the original time points, and evaluates the approach on a TSSB-derived inductive benchmark built from disjoint UCR training and test instances.
Ivana Kesić, Carolina Fortuna, M. Mohorčič et al.· arXiv.org· 0 citations
This work introduces agentic-eCAL, generalizing the Energy Cost of AI Lifecycle (eCAL) metric to directed multi-agent workflows by coupling a closed-form two-rate single-call energy model (compute-bound prefill and memory-bound decode) with 7-layer OSI data transport.
Carolina Fortuna, Vid Hanžel, Tim Strnad et al.· 0 citations
MKAN is competitive with state-of-the-art monotone NNs on the SMM/ICML-2024 benchmark while being the only method that combines hard unconstrained monotonicity with KAN's per-edge functional transparency.
