AI & ML 2025-12-16
Nemotron 3 Nano sets a new benchmark for efficient, open agentic models. CUGA on Hugging Face advances configurable AI agents, while NVIDIA acquires SchedMD to enhance workload management. New guides enable streamlined LLM fine-tuning on NVIDIA GPUs.
Nemotron 3 Nano - A new Standard for Efficient, Open, and Intelligent Agentic Models 92
Tags:
大模型多智能体系统混合架构长上下文开源模型推理效率Source:
HuggingFace| 阅读原文
[摘要]
Nemotron 3 Nano 30B提出一种混合Mamba-Transformer MoE架构,实现1M上下文窗口下的高效高精度推理,专为多智能体系统设计,兼具低延迟、高吞吐与开放可复现性。
CUGA on Hugging Face: Democratizing Configurable AI Agents 92
Tags:
AI AgentConfigurable AgentMulti-tool IntegrationLangChainHugging Face SpacesSource:
HuggingFace| 阅读原文
[摘要]
CUGA提出一种可配置的通用AI代理框架,通过结构化规划与多工具集成实现跨域复杂任务的高可靠执行,在AppWorld和WebArena等基准上达顶尖性能。
NVIDIA Acquires Open-Source Workload Management Provider SchedMD 85
Tags:
SlurmHPCAI infrastructureworkload managementopen-sourceresource schedulingSource:
NVIDIA_Blog| 阅读原文
[摘要]
NVIDIA收购Slurm开发商SchedMD,强化其在HPC与AI领域的开源工作负载管理能力,推动跨硬件平台的高效资源调度与规模化AI训练。
How to Fine-Tune an LLM on NVIDIA GPUs With Unsloth 85
Tags:
参数高效微调小模型优化代理AILoRAQLoRANVIDIA NemotronSource:
NVIDIA_Blog| 阅读原文
[摘要]
提出高效参数化微调(如LoRA/QLoRA)与全量微调结合的实践框架,推动小模型在专用代理任务中实现高精度、低资源部署。