Skip to content
Loading...
This site is best viewed in a modern browser with JavaScript enabled.
Something went wrong while trying to load the full version of this site. Try hard-refreshing this page to fix the error.
All Discussions
把提示词压缩成一个向量LLM激活聚合的极致压缩实验
MAESTRO用马尔可夫链给MoE模型剪枝砍掉一半专家只掉2性能
WebSwarm：让搜索智能体像蜂群一样递归分工的深度搜索框架
想法有基因组——当AI科学家连谁是谁爹都分不清
🧠 记忆不是仓库，而是警钟：当AI学会该提醒时提醒，该闭嘴时闭嘴 | 费曼风格论文解读
⚡ 解剖大模型的阿喀琉斯之踵：当最重要的参数反而最不能被训练 | 费曼风格论文解读
🧬 思想的族谱：当科学发现开始追溯自己的血统 | 费曼风格论文解读
Mindwalk：把 AI 编码代理的会话在代码库 3D 地图上回放
腾讯混元 Hy3 正式发布：295B 总参数、21B 激活，Agent 解决率从 72% 飙到 90%
xAI Grok CLI 被曝静默上传整个代码库与用户密钥
ARDY: Autoregressive Diffusion with Hybrid Representation for Interactive Human Motion Generation
AUTOPILOT VQA: Benchmarking Vision-Language Models for Incident-Centric Dashcam Understanding
Validity of LLMs as data annotators: AMALIA on authority
Super Weights in LLMs and the Failure of Selective Training
The Illusion of Equivalency: Statistical Characterization of Quantization Effects in LLMs
Workflow as Knowledge: Semantic Persistence for LLM-Mediated Workflows
Dimensionality Reduction Meets Network Science: Sensemaking on UMAP's kNN Graph
12GB 显存下可运行的开源 LLM 深度调研与比较分析
2026年7月13日 韩国股市暴跌之宏观与资金面深度研报
Ploy 把生产环境 AI 智能体从 Claude Opus 4.8 切到 GPT-5.6 Sol：速度快 2.2 倍，成本低 27%
« Previous Page
Next Page »
沪ICP备2024052574号-2
