Harnesses for large language models have converged on a single-threaded shape: one agent, one history, one machine, one sitting. That shape fails work that spans many agents, outlives any one model call, or waits on systems that are not agents. We describe the Agent Board, the coordination substrate in the Corvic harne...
Tyler Hunt, Donald Nguyen, Gurbinder Gill· 0 citations
Real-world problems span diverse domains, including program synthesis, symbolic reasoning, planning, and optimization, and often require fundamentally different solution paradigms. A central challenge for LLM-based reasoning is twofold: identifying an appropriate problem-solving approach and formulating it correctly. I...
Yilun Hao, Ruixiao Yang, Hudson Hilal et al.· 0 citations
It is found that although hard problems lead to modest gains overall, they are actually the ones that help most in expanding the model's reasoning capabilities and pushing the representations, which helps to better understand some of the prior conflicting observations about RLVR.
Phuc Minh Nguyen, Ngoc-Hieu Nguyen, P. Shojaee et al.· 0 citations
As large language models are fine-tuned for increasingly specialized tasks, their performance depends more on which training examples are selected than on how many. Given a large heterogeneous training pool and a small target set specifying the desired capability, targeted data selection aims to find a compact subset t...
Zheng-Hao Zhao, Yasutomo Kawanishi, Hui-Huo Zheng et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
In mixed-vendor accelerator fleets, many training jobs can run on several hardware types, and the best choice depends on each job's throughput on each type. Routing a job needs a throughput estimate for every candidate type before the job runs. We present LEAP, a machine learning performance prediction service that pre...
Hossein Askari, Wei Sun, Chenguang Zhu et al.· 0 citations
In speculative decoding, parallel drafters such as DFlash make drafting cheap, so target verification dominates each round. A central choice is the verification length, the number of tokens that the target processes in each round. Verifying more draft tokens can accept more of them, but work after the first rejection i...
Haiyu Wang, Sai Qian Zhang, Zechun Liu et al.· 0 citations
The uncertainty of a statistical model is most commonly factorized into an aleatoric and an epistemic part. This factorization changes how predictions are interpreted in downstream decision tasks. Importantly, except for the idealistic scenario with no model mismatch, the quantification is a characteristic of the model...
Manuel Wendl, Erik Englesson, Andreas Krause et al.· 0 citations
Large Language Models (LLMs) serve both as tools for aiding human decision-making in everyday life and as models of human agents in experimental simulation studies. Alignment on prototypical social decisions -- such as whether to cooperate on a collective action -- therefore has immediate practical consequences for res...
Leon Klingborg, Theodor Cimpeanu, Alexander J. Stewart· 0 citations
Foundation models in changing environments can encounter unfamiliar domain knowledge that their original training did not cover. However, new domains rarely provide large amounts of expert-created training data or reliable supervisory signals for further training. We introduce Recursa, a foundation model system that ad...
Erkang Zhu, Yin Lin, Xiaorong Shi et al.· 0 citations
A five-candidate short list contains a deployment within 12% of the best feasible pairing of model and latency target, and a five-candidate short list contains a deployment within 12% of the best feasible pairing of model and latency target.
Anshu Raina, Yao Fu, Jie-Jing Zhang et al.· 0 citations
AdaIE, a parameter-free mechanism that reweights pretrained representations according to the evolving downstream sensitivity during training, is introduced, highlighting interface co-adaptation as an important optimization problem when coupling pretrained representations with downstream models.
Rui-Jie Zhang, Jia-Yin Zheng, Ye-Quan Zhao et al.· 0 citations
WASP-Mem (Wiki-based Agent-Synthesized Procedural Memory), a nimble but effective method that distills execution traces and feedback into persistent, interlinked wiki pages, enabling continual agent learning is introduced.
Pei-Zhi Niu, Ge Shi, Vaibhav Gorde et al.· 0 citations
