2026-11-17 周二 · 生成式推荐
Generative生成式推荐 · 1 篇Personalization个性化推荐 · 2 篇AI SearchAI 搜索 · 10 篇Infra系统与索引 · 10 篇Embodied AI具身智能 · 10 篇
今日场景今天的生成式赛道只有一篇,问的是长历史到底能不能被压进固定容量的记忆里,还不把预测需要的信号丢掉。生成式推荐一旦把很长的交互直接送进模型,计算和显存会跟着长度一起涨。这篇把历史分段写入数量固定的记忆槽,解码时只用固定长度的上下文,并在电商长序列基准上表明,压缩后仍能接近直接处理长历史的对照模型,远期行为也还带得来信号。对要上线长序列生成式推荐的人来说,关键不是把注意力窗口无限拉长,而是让记忆在历史不断变长时仍然稳定可用。 — 评分 13 篇,入选 1 篇,分数 4.30 – 4.30
LEAD STORY · 今日必读 ·
Alibaba /
Shanghai Jiao Tong University
RecMem: Recurrent Memory Compression for Long-Sequence Generative Recommendation
生成式推荐一旦把用户很长的历史直接送进模型,计算量和显存就会跟着历史长度一起涨,很长的交互序列很快变得又慢又贵。作者把历史一段一段压进数量固定的记忆槽,解码时只用固定长度的上下文,不再随历史变长而膨胀。在来自电商平台的长序列基准上,解码上下文从 1200 个 token 减到 216 个,仍保住了直接处理同样长历史的对照模型 87.9% 的 Recall@10,更靠后的召回也几乎相当。
4.30
综合分 / 5
偏好调整 4.30
偏好调整 4.30
Haoyi Hu, Gaoyang Guo, Yunjia Xi, Wen Chen, Weiwen Liu, Jian Wu, Yuning Jiang 等 8 人 · 2026-11-13 · arXiv
其他赛道今日
点击切换开源雷达
GitHub / HuggingFace 热门 · 8 个项目AayushPandey2/teachme-skillAn AI agent skill (/teachme) that deeply researches any topic, builds a complete Markdown study vault, and teaches it to you step by step.★ 10 · 2026-11-15 · GitHubsunkpoet-lang/SUNSET-RIDERS-RECOMPExperimental Sunset Riders PC port and recompilation research. Four-player local gameplay; static recompilation and online support are works in progress. No ROMs included..C++ · ★ 10 · 2026-11-13 · GitHubcjryrx/a7c2-precaptureA7C II 800ms precapture research patch tools. Source only, no firmware; non-commercial research use.Python · ★ 9 · 2026-11-11 · GitHubinternlm/WildClawBench-TrajectoriesWildClawBench Trajectories What 12 frontier models did on 60 real-world agent tasks, step by step, with the score of every run 720 graded trajectories. 12 models. 15,678 model steps. 17,206 tool calls. 90 hours of agent time. Every model on this list ran the full WildClawBench su; tags: task_categories:text-generation, language:en, language:zh, license:mit, size_categories:n<1K, format:json, modality:tabular, modality:textdataset · ★ 8 · 2026-07-30 · HuggingFacedeepseek-ai/DeepSeek-V4.1-Flash-1017pipeline: image-text-to-text; tags: transformers, safetensors, deepseek_v41, text-generation, image-text-to-text, conversational, license:mit, eval-resultsmodel · ★ 5785 · 2026-10-17 · HuggingFaceinternlm/WildClawBenchWildClawBench Hard, practical, end-to-end evaluation for AI agents — in the wild. WildClawBench is an agent benchmark that tests what actually matters: can an AI agent do real work, end-to-end, without hand-holding? We drop agents into a live OpenClaw environment — the same open-; tags: benchmark:official, task_categories:visual-question-answering, task_categories:image-text-to-text, task_categories:question-answering, language:en, language:zh, license:mit, size_categories:n<1Kdataset · ★ 73 · 2026-03-24 · HuggingFaceinternlm/Sage-S2.1-Flashtags: safetensors, glm5_next, region:usmodel · ★ 46 · 2026-11-15 · HuggingFaceinternlm/Sage-S2.1-Flash-FP8tags: safetensors, glm5_next, fp8, region:usmodel · ★ 5 · 2026-11-13 · HuggingFace