谷歌发布手语转文本模型 SL2T

2026-08-13 · 共 30 条候选

2026-08-13 共收录 30 条候选新闻，本期为兜底版面。以下为各条目详情：【GoogleDeepMind】谷歌发布手语转文本模型 SL2T。Google DeepMind 推出 SL2T，可将美国手语实时翻译为英语，并集成到 Pixel 11 的 Gboard 与 Live Transcribe。 原推文：SL2T is our breakthrough sign language-to-text model powering new features for Deaf and hard of hearing users on @Android. Starting with American Sign Language-to-English on Pixel 11, people can sign directly into Gboard and Live Transcribe instead of typing. https://t.co/p9Vx7tJtLT 数据：点赞 1065，转推 152，浏览 82827。　【huggingface】NVIDIA 开源 Nemotron 3.5 Lightning。30B MoE 模型，3B 激活参数，面向 always-on Agent，权重、数据与训练配方全部开放。 原推文：RT @NVIDIAAI: As always, NVIDIA Nemotron 3.5 Lightning is open and customizable. This includes weights, data and recipes. Available now on… 数据：点赞 0，转推 20，浏览 2191。　【TencentHunyuan】腾讯发布自进化智能体可靠性综述。论文系统梳理智能体自我进化方法，并探讨如何验证每次更新真正带来改进。 原推文：📄New Research on Self-Evolving Agents: When AI agents modify themselves, how do we know they actually got better? We present Diving into Reliable Self-Evolving Agents: A Survey—a systematic map of how agents self-evolve and what evidence is needed to trust each update. The https://t.co/eIinsfhebv 数据：点赞 333，转推 42，浏览 17917。　【OpenAIDevs】OpenAI Codex 桌面版正式支持 Linux。Codex 桌面版现可在 Ubuntu、Debian 和 Fedora 上使用，支持并行 Agent。 原推文：RT @reach_vb: yesterday we shipped Codex for Linux 🐧 the full Codex desktop experience is now on Ubuntu, Debian and Fedora parallel agent… 数据：点赞 0，转推 13，浏览 4352。　【Alibaba_Qwen】Qwen3.8-Max 法律推理榜升至第四。阿里 Qwen3.8-Max 在 Legal Research Bench 的排名从第 22 跃升至第 4。 原推文：From #22 to #4 on Legal Research Bench! Solid progress for Qwen3.8-Max. Thanks for highlighting~✨ 数据：点赞 638，转推 34，浏览 58311。　【huggingface】MOSS-VL 发布 FP8/NF4 量化版。24GB 显存即可本地运行 MOSS-VL，官方推出 FP8 与 NF4 版本。 原推文：RT @MosiAI_Official: 24GB VRAM is enough to run MOSS-VL locally.@Open_MOSS We’ve released FP8 and NF4 versions for both MOSS-VL-Instruct a… 数据：点赞 0，转推 11，浏览 2250。　【cohere】Today, we’re adding another me。Today, we’re adding another member to our model family. Meet North Micro Vision. Our smallest visio 原推文：Today, we’re adding another member to our model family. Meet North Micro Vision. Our smallest vision-language model yet, ideal for sophisticated document understanding. Available open-source under an Apache 2.0 license. Get the weights on @huggingface. https://t.co/dc7jz8UpUo 数据：点赞 497，转推 55，浏览 36435。　【nvidia】Congrats to the @SpaceXAI team。Congrats to the @SpaceXAI team on the release of Grok 4.6. Grok 4.6 brings frontier intelligence, r 原推文：Congrats to the @SpaceXAI team on the release of Grok 4.6. Grok 4.6 brings frontier intelligence, running and trained on NVIDIA GB300 NVL72 with NVLink to deliver exceptional performance, reliability and lowest token cost. 数据：点赞 6076，转推 540，浏览 455594。　【nvidia】We’re honored that Glassdoor h。We’re honored that Glassdoor has named our founder and CEO, @JensenHuang , No. 1 on its 2026 Best CE 原推文：We’re honored that Glassdoor has named our founder and CEO, @JensenHuang , No. 1 on its 2026 Best CEOs list. 🎉 A very meaningful recognition, grounded in feedback from the people who make NVIDIA what it is. https://t.co/At3CEoP5ji 数据：点赞 456，转推 58，浏览 51243。　【runwayml】Grok Imagine Image 2.0 is now 。Grok Imagine Image 2.0 is now on Runway. The newest model available now, alongside the world's best 原推文：Grok Imagine Image 2.0 is now on Runway. The newest model available now, alongside the world's best image and video models. Try it now at the link below. https://t.co/Y19jBujPcs 数据：点赞 21，转推 4，浏览 3610。　【runwayml】You can now connect Figma, Dro。You can now connect Figma, Dropbox and Notion to Agent on Runway. Your assets shouldn't be stuck in 原推文：You can now connect Figma, Dropbox and Notion to Agent on Runway. Your assets shouldn't be stuck in separate tools. With connectors, Agent syncs directly with the platforms you already work in, pulling your designs, files and docs into one place. Available on all paid plans. https://t.co/0AGN7iteJc 数据：点赞 57，转推 5，浏览 8054。　【runwayml】LTX-2.5 is now on Runway. Avai。LTX-2.5 is now on Runway. Available today. Get started at the link below. https://t.co/fOKZ63ZwBY 原推文：LTX-2.5 is now on Runway. Available today. Get started at the link below. https://t.co/fOKZ63ZwBY 数据：点赞 143，转推 8，浏览 12118。　【LumaLabsAI】FLUX 3 is now live in Luma. @。FLUX 3 is now live in Luma. @bfl_ai 's first general model for image, video, audio, and action-pred 原推文：FLUX 3 is now live in Luma. @bfl_ai 's first general model for image, video, audio, and action-prediction. Video with native audio, available today. More coming soon. https://t.co/fRUgGAc8Yr 数据：点赞 57，转推 3，浏览 4988。　【cursor_ai】RT @SpaceXAI: Introducing Grok。RT @SpaceXAI: Introducing Grok 4.6. It delivers frontier intelligence and is a significant improvem 原推文：RT @SpaceXAI: Introducing Grok 4.6. It delivers frontier intelligence and is a significant improvement over Grok 4.5 at the same price. ht… 数据：点赞 0，转推 2807，浏览 4534。　【vercel】We built a software factory fo。Vercel 为 AI SDK 构建了一个软件工厂，自动化处理 Issue 和 PR，但所有合并仍由人工控制。运行四周后，该工厂参与了 25%–40% 合并 PR 的编写；7 月关闭了 70% 的 Issue；未解决的 Bug 数量下降 25%。 原推文：We built a software factory for AI SDK. Each step is an agent, and humans merge changes. Four weeks in: ▪️ The factory authors up to 35% of merged PRs ▪️ It closed 70% of issues in July ▪️ Open bugs are down 25% https://t.co/UNGeeQJdAu 数据：点赞 566，转推 28，浏览 119474。　【vercel】We migrated the database behin。Vercel 将支撑每次构建的“构建热池”数据库从 Redis 迁移到 DynamoDB，并在每分钟处理 6000 次部署的生产环境中完成切换。迁移采用功能开关分阶段推进，每个阶段都可回滚。博客讲述了迁移过程、工程理念及经验教训。 原推文：We migrated the database behind every build at Vercel, one of our most critical services, while it handled 6,000 deployments a minute. The story of how it happened, our philosophy, and what we learned ↓ https://t.co/vhevq1B4Vx 数据：点赞 232，转推 13，浏览 62679。　【replit】Want lower latency in Replit, 。Want lower latency in Replit, and control over where your project previews are hosted? Pro & En 原推文：Want lower latency in Replit, and control over where your project previews are hosted? Pro & Enterprise customers can now choose where new Replit workspaces are hosted. Pick between North America, Europe, or Asia. https://t.co/VJHInqPJo2 数据：点赞 46，转推 3，浏览 6246。　【llama_index】RT @jerryjliu0: We wrote a 36-。RT @jerryjliu0: We wrote a 36-page ArXiv whitepaper on ExtractBench 🧑‍🔬 , our effort to create the 原推文：RT @jerryjliu0: We wrote a 36-page ArXiv whitepaper on ExtractBench 🧑‍🔬 , our effort to create the most comprehensive, schema-guided, real-… 数据：点赞 0，转推 9，浏览 307。　【llama_index】The most dangerous document ex。The most dangerous document extraction failure isn't a wrong value. It's a missing row that looks li 原推文：The most dangerous document extraction failure isn't a wrong value. It's a missing row that looks like nothing is wrong. We released ExtractBench yesterday: 370 enterprise docs, 14 systems. The hardest test: long-list completeness. An unclaimed-property list with 26,725 rows. A https://t.co/NgY4ZHFirf 数据：点赞 19，转推 4，浏览 2806。　【llama_index】RT @jerryjliu0: ExtractBench i。RT @jerryjliu0: ExtractBench is one of the most comprehensive benchmarks for real-world document ext 原推文：RT @jerryjliu0: ExtractBench is one of the most comprehensive benchmarks for real-world document extraction. ✅ It covers 4869 pages, acro… 数据：点赞 0，转推 12，浏览 448。　【ollama】RT @burkeholland: Qwen3-Coder 。RT @burkeholland: Qwen3-Coder 30B running locally in the @GitHub Copilot app (@ollama) spitting ~85 原推文：RT @burkeholland: Qwen3-Coder 30B running locally in the @GitHub Copilot app (@ollama) spitting ~85 tokens per second on 96GB VRAM. It's… 数据：点赞 0，转推 5，浏览 1143。　【ollama】You can now use Ollama as a pr。GitHub Copilot for JetBrains 更新，新增 Ollama 作为模型提供方，支持接入本地模型。此次更新还带来持久记忆功能，增强企业控制，改进日常聊天工作流，并修复了 MCP 服务器的可靠性问题。 原推文：You can now use Ollama as a provider in GitHub Copilot for JetBrains. https://t.co/Rt8iSMYWaA https://t.co/SGZZ8VGCDU 数据：点赞 175，转推 17，浏览 13683。　【higgsfield_ai】Decomposing a flat poster like。Decomposing a flat poster like this used to take hours. Higgsfield Layers automatically separates h 原推文：Decomposing a flat poster like this used to take hours. Higgsfield Layers automatically separates headlines, graphic badges, and masked characters into isolated layers. Production-ready layer decomposition in seconds. Higgsfield Layers is live. https://t.co/9KdZcACis4 数据：点赞 103，转推 19，浏览 17855。　【synthesiaIO】Most teams slow down in summer。Most teams slow down in summer. We accelerated 🚢💨 Here’s what we shipped in July. https://t.co/ 原推文：Most teams slow down in summer. We accelerated 🚢💨 Here’s what we shipped in July. https://t.co/I7vmBEsb29 数据：点赞 6，转推 3，浏览 984。　【recraftai】Seedance 2.5 is now live in Re。Seedance 2.5 is now live in Recraft Studio. Generate videos up to 30 seconds long, no cuts needed. 原推文：Seedance 2.5 is now live in Recraft Studio. Generate videos up to 30 seconds long, no cuts needed. Check it out today! https://t.co/ySeMeZAh4W 数据：点赞 4，转推 0，浏览 593。　【AdobeFirefly】Get you an AI animation genera。Adobe Firefly 推出免费在线 AI 动画生成器，可将用户想法和角色快速转化为 2D 与 3D 动画内容。该工具强调兼顾专业级动画质量和商业可用性，适合将创意直接生成可安全用于商业场景的动画作品。 原推文：Get you an AI animation generator who can do both: professional quality animations *and* commercially-safe results. Bring your ideas to life with an online animation generator: https://t.co/GysEJiioTR https://t.co/W90cEX5Acg 数据：点赞 5，转推 1，浏览 194。　【Kling_ai】Have you heard? This little pi。Have you heard? This little pirate cat is going viral worldwide 🐱 Full video drops tomorrow — stay 原推文：Have you heard? This little pirate cat is going viral worldwide 🐱 Full video drops tomorrow — stay tuned! https://t.co/sQsZReYVH8 数据：点赞 35，转推 0，浏览 12325。　【Hailuo_AI】RT @aimikoda: MiniMax H3 + Sun。RT @aimikoda: MiniMax H3 + Suno Beat-Synced Character Showcase Across 5 Environments First create 原推文：RT @aimikoda: MiniMax H3 + Suno Beat-Synced Character Showcase Across 5 Environments First create a 15-seconds Suno track with style "pun… 数据：点赞 0，转推 35，浏览 652。　【Hailuo_AI】RT @aimikoda: Player stats UI 。RT @aimikoda: Player stats UI animation with GPT Image 2 and MiniMax H3. This model is amazing for 原推文：RT @aimikoda: Player stats UI animation with GPT Image 2 and MiniMax H3. This model is amazing for UI animations. Give GPT Image 2 your ch… 数据：点赞 0，转推 65，浏览 618。　【Hailuo_AI】RT @Jinghui_Dong: Start to exp。RT @Jinghui_Dong: Start to explore 3rd person shooting game simulation by using Minimax H3. Here is 原推文：RT @Jinghui_Dong: Start to explore 3rd person shooting game simulation by using Minimax H3. Here is what I have found: 01. A lot of fun.… 数据：点赞 0，转推 4，浏览 683。

本期兜底版面由系统自动生成，内容来源于 AI 领域各账号当日发布内容，经过评分筛选后按重要性排列。如需完整版面，请等待次日更新。本页所有内容均经过自动抓取与处理，仅供参考，不代表本站立场。如有内容错误或版权问题，请通过官方渠道联系处理。感谢您的理解与支持。数据来源：Twitter/X 公开时间线，仅展示公开发布的原创内容。版权归原作者所有，转载请注明出处。兜底版面在自动排版管线异常时启用，确保读者每日仍可获取经过筛选与评分的核心条目，避免出现空白页面。日报覆盖模型发布、产品更新、研究论文、行业动态、个人观点等主题，按重要性自动归类与编排。更详细的版面、配图与多语言版本将在管线恢复后于次日补齐，敬请关注。