On Federated Averaging with Partial Participation under a General Variance Condition
arXiv:2611.14239v1 Announce Type: cross Abstract: In this paper, we study the convergence of Federated Averaging (FedAvg) with partial participation under a gen…
言栖资讯收录 250809 篇 AI、开发与建站文章,按发布时间排序,支持搜索与按语言筛选。
arXiv:2611.14239v1 Announce Type: cross Abstract: In this paper, we study the convergence of Federated Averaging (FedAvg) with partial participation under a gen…
arXiv:2611.13730v1 Announce Type: cross Abstract: Optimized floating-point inference is not reproducible across environments: the same model on different hardwa…
arXiv:2611.13557v1 Announce Type: cross Abstract: We present Darmok, the first automated system to reach superhuman performance on British cryptic crosswords. U…
arXiv:2611.12961v1 Announce Type: cross Abstract: Federated fine-tuning with low-rank adaptation (LoRA) reduces communication and computational overhead by trai…
arXiv:2611.14204v1 Announce Type: new Abstract: Scientific computing is under increasing pressure to adapt to HPC hardware that is shaped by AI workloads. In th…
arXiv:2611.14051v1 Announce Type: new Abstract: Reusing key--value (KV) caches across language model updates could avoid repeated computation of shared context,…
arXiv:2611.14026v1 Announce Type: new Abstract: Serverless computing promises to run code only on demand, yet the servers beneath it draw most of their power wh…
arXiv:2611.13802v1 Announce Type: new Abstract: LLM services have become essential infrastructure for applications such as chat assistants and coding agents. Se…
arXiv:2611.13740v1 Announce Type: new Abstract: Large language models (LLMs) have been increasingly composed into multi-agent workflows to accomplish complex an…
arXiv:2611.13442v1 Announce Type: new Abstract: Parallel drafters accelerate LLM decoding by proposing a whole block of draft tokens in a single forward pass, m…
arXiv:2611.13413v1 Announce Type: new Abstract: DeepSeek-V4.1's 203 GB of Engram n-gram tables sit in host DRAM today, taking 190 of the 223 GiB a 4-GPU serving…
arXiv:2611.13412v1 Announce Type: new Abstract: LLM agents drive inference engines with closed-loop sessions whose context grows every turn, and serving them ec…
arXiv:2611.13393v1 Announce Type: new Abstract: Correctness is a prerequisite for training large language models (LLMs) at scale, yet it is hard to ensure: a fa…
arXiv:2611.13309v1 Announce Type: new Abstract: AI coding agents can now construct scientific workflows, but operating them on shared cyberinfrastructure (CI) r…
arXiv:2611.13293v1 Announce Type: new Abstract: Serving large language models at high concurrency with long contexts requires KV-cache capacity and attention ba…
arXiv:2611.13188v1 Announce Type: new Abstract: Expert parallelism (EP) shards the experts of a Mixture-of-Experts (MoE) layer across GPUs, so every MoE layer r…
arXiv:2611.13176v1 Announce Type: new Abstract: Per-step model routers send each LLM call of an agent to the cost-optimal model dynamically within each session.…
arXiv:2611.13019v1 Announce Type: new Abstract: The proliferation of maritime activities has spurred the emergence of numerous computation-intensive and delay-s…
arXiv:2611.10035v2 Announce Type: replace-cross Abstract: End-to-end VITS text-to-speech models synthesize natural speech in a single forward pass, yet standard…
arXiv:2611.10693v2 Announce Type: replace Abstract: Multimodal LLMs have made remarkable progress in perceiving the physical world, while their perceptual scope…