<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[Together_CZ的博客]]></title><description><![CDATA[种一棵树，最好的时间是十年前，其次是现在]]></description><link>https://blog.youkuaiyun.com/Together_CZ</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; Together_CZ]]></copyright><item><title><![CDATA[NeMo Data Designer: An Extensible Framework for Multimodal Synthetic Data Generation一个用于多模态合成数据生成的可扩]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/168132306</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/168132306</guid><author>Together_CZ</author><pubDate>Thu, 12 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[NeMo Data Designer: An Extensible Framework for Multimodal Synthetic Data Generation——一个用于多模态合成数据生成的可扩展框架]]></description><category></category></item><item><title><![CDATA[ToolQA: A Dataset for LLM Question Answering with External Tools——一个用于评估大语言模型使用外部工具进行问答的数据集]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/170052054</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/170052054</guid><author>Together_CZ</author><pubDate>Wed, 11 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[ToolQA: A Dataset for LLM Question Answering with External Tools——一个用于评估大语言模型使用外部工具进行问答的数据集]]></description><category></category></item><item><title><![CDATA[Clearing the FoG: Training KG Agents to Retrieve the Evidence Behind Their Answers拨开迷雾：训练知识图谱智能体检索其答]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/170051756</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/170051756</guid><author>Together_CZ</author><pubDate>Tue, 10 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[Clearing the FoG: Training KG Agents to Retrieve the Evidence Behind Their Answers——拨开迷雾：训练知识图谱智能体检索其答案背后的证据]]></description><category></category></item><item><title><![CDATA[SEAKR: Self-aware Knowledge Retrieval for Adaptive Retrieval Augmented Generation面向自适应检索增强生成的自感知知识检索]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/169913935</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/169913935</guid><author>Together_CZ</author><pubDate>Mon, 09 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[SEAKR: Self-aware Knowledge Retrieval for Adaptive Retrieval Augmented Generation——面向自适应检索增强生成的自感知知识检索]]></description><category></category></item><item><title><![CDATA[Large Knowledge Model: A Knowledge Foundation for Agentic Science at Scale——大知识模型：面向规模化智能体科学的知识基础]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/169843853</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/169843853</guid><author>Together_CZ</author><pubDate>Sun, 08 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[Large Knowledge Model: A Knowledge Foundation for Agentic Science at Scale——大知识模型：面向规模化智能体科学的知识基础]]></description><category></category></item><item><title><![CDATA[EvoSkill: Automated Skill Discovery for Coding Agents——面向多智能体系统的自动化技能发现]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/168022063</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/168022063</guid><author>Together_CZ</author><pubDate>Sat, 07 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[EvoSkill: Automated Skill Discovery for Coding Agents——面向多智能体系统的自动化技能发现]]></description><category></category></item><item><title><![CDATA[Trace2Skill: Distill Trajectory-Local Lessons into Transferable Agent Skills——将轨迹局部经验提炼为可迁移的智能体技能]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/168011483</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/168011483</guid><author>Together_CZ</author><pubDate>Fri, 06 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[Trace2Skill: Distill Trajectory-Local Lessons into Transferable Agent Skills——将轨迹局部经验提炼为可迁移的智能体技能]]></description><category></category></item><item><title><![CDATA[SkillOptLite and HarnessOpt: Better and Faster Agent Self-evolution with One Line of Vibe用一行 Vibe 实现]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/168018493</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/168018493</guid><author>Together_CZ</author><pubDate>Thu, 05 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[SkillOptLite and HarnessOpt: Better and Faster Agent Self-evolution with One Line of Vibe——用一行 Vibe 实现更好、更快的智能体自我进化]]></description><category></category></item><item><title><![CDATA[Taramandal-GPT: Enhancing Astrodynamics Problem-Solving with Knowledge Retrieval and Structured Thin]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/169459815</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/169459815</guid><author>Together_CZ</author><pubDate>Wed, 04 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[Taramandal-GPT: Enhancing Astrodynamics Problem-Solving with Knowledge Retrieval and Structured Thinking——通过知识检索与结构化思维增强天体动力学问题求解]]></description><category></category></item><item><title><![CDATA[NumericJev: Jev-like LLM Numerical Decoding with Multiway Decision Trees——类Jev大语言模型基于多路决策树的数值解码]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/169365527</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/169365527</guid><author>Together_CZ</author><pubDate>Tue, 03 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[NumericJev: Jev-like LLM Numerical Decoding with Multiway Decision Trees——类Jev大语言模型基于多路决策树的数值解码]]></description><category></category></item><item><title><![CDATA[AnyChange ：Segment Any Change——分割任意变化]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/167802676</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/167802676</guid><author>Together_CZ</author><pubDate>Mon, 02 Nov 2026 01:00:00 +0800</pubDate><description><![CDATA[AnyChange ：Segment Any Change——分割任意变化]]></description><category></category></item><item><title><![CDATA[From human experts to machines: An LLM supported approach to ontology and knowledge graph constructi]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/169061684</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/169061684</guid><author>Together_CZ</author><pubDate>Sat, 31 Oct 2026 01:00:00 +0800</pubDate><description><![CDATA[From human experts to machines: An LLM supported approach to ontology and knowledge graph construction ⋆——从人类专家到机器]]></description><category></category></item><item><title><![CDATA[VIPS: a Vision-based Page Segmentation Algorithm——一种基于视觉的页面分割算法]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/169461612</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/169461612</guid><author>Together_CZ</author><pubDate>Fri, 30 Oct 2026 01:00:00 +0800</pubDate><description><![CDATA[VIPS: a Vision-based Page Segmentation Algorithm——一种基于视觉的页面分割算法]]></description><category></category></item><item><title><![CDATA[Towards the Automation of Knowledge Graph Construction Using Large Language Models——迈向知识图谱的自动化]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/169061350</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/169061350</guid><author>Together_CZ</author><pubDate>Fri, 30 Oct 2026 01:00:00 +0800</pubDate><description><![CDATA[Towards the Automation of Knowledge Graph Construction Using Large Language Models——迈向知识图谱的自动化]]></description><category></category></item><item><title><![CDATA[ModularRSI: Modular and Generalizable Recursive Harness Self-Improvement——模块化且可泛化的递归式脚手架自我改进]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/167909906</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/167909906</guid><author>Together_CZ</author><pubDate>Fri, 30 Oct 2026 01:00:00 +0800</pubDate><description><![CDATA[ModularRSI: Modular and Generalizable Recursive Harness Self-Improvement——模块化且可泛化的递归式脚手架自我改进]]></description><category></category></item><item><title><![CDATA[SAG2: Embedding-Free Multi-Hop Retrieval with BM25 and JEV——基于 BM25 和 JEV 的无嵌入多跳检索]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/168873752</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/168873752</guid><author>Together_CZ</author><pubDate>Thu, 29 Oct 2026 01:00:00 +0800</pubDate><description><![CDATA[SAG2: Embedding-Free Multi-Hop Retrieval with BM25 and JEV——基于 BM25 和 JEV 的无嵌入多跳检索]]></description><category></category></item><item><title><![CDATA[CMB: A Coding Memory Benchmark for Long-Horizon and Interleaved Coding Agents——面向长程与交错式编码智能体的编码记忆基准]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/168246636</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/168246636</guid><author>Together_CZ</author><pubDate>Wed, 28 Oct 2026 01:00:00 +0800</pubDate><description><![CDATA[CMB: A Coding Memory Benchmark for Long-Horizon and Interleaved Coding Agents——面向长程与交错式编码智能体的编码记忆基准]]></description><category></category></item><item><title><![CDATA[Has GPT-6-Astra Achieved Embodied Intelligence?——GPT-6-Astra 是否实现了具身智能？]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/168658457</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/168658457</guid><author>Together_CZ</author><pubDate>Tue, 27 Oct 2026 01:00:00 +0800</pubDate><description><![CDATA[Has GPT-6-Astra Achieved Embodied Intelligence?——GPT-6-Astra 是否实现了具身智能？]]></description><category></category></item><item><title><![CDATA[SwarmRSI: Recursive Self-Improvement through Evolving Scientific Agent Swarm——通过演化科学智能体群体实现递归自我改进]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/168549367</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/168549367</guid><author>Together_CZ</author><pubDate>Mon, 26 Oct 2026 01:00:00 +0800</pubDate><description><![CDATA[SwarmRSI: Recursive Self-Improvement through Evolving Scientific Agent Swarm——通过演化科学智能体群体实现递归自我改进]]></description><category></category></item><item><title><![CDATA[世界模型技术发展与趋势演变研究与分析]]></title><link>https://blog.youkuaiyun.com/Together_CZ/article/details/169041649</link><guid>https://blog.youkuaiyun.com/Together_CZ/article/details/169041649</guid><author>Together_CZ</author><pubDate>Sun, 25 Oct 2026 08:53:34 +0800</pubDate><description><![CDATA[世界模型是智能体学习环境动态并预测未来状态以支持规划与决策的核心工具，不等同于视频生成或传统仿真器。其核心价值在于对动作条件的响应能力与物理一致性，而非视觉逼真度。当前技术呈现多路线融合趋势，涵盖潜在状态预测、3D空间建模、行动-结果联合建模等。未来将从“生成可信画面”转向“可控、可信赖的未来推演”，更可能形成由预训练模型、动力学模块与规划系统组成的分层闭环体系。评估应聚焦任务实效，而非仅看生成效果。]]></description><category></category></item></channel></rss>