liuyong @forrestbing
I am a researcher in AIGC, Multi-modality and VitrualHuman tech direction 中华人民共和国 Joined April 2016-
Tweets317
-
Followers242
-
Following5K
-
Likes943
Anthropic 这个教程教你如何创建一个自己的语言模型评估测试集。 1️⃣LLM 评估 体系通常包含4个部分: 输入提示集 模型对这些提示的响应 用来与模型输出对比的“标准答案” 根据某种评分方法得出的分数 2️⃣前三部分相当直观 ——…
Anthropic 这个教程教你如何创建一个自己的语言模型评估测试集。 1️⃣LLM 评估 体系通常包含4个部分: 输入提示集 模型对这些提示的响应 用来与模型输出对比的“标准答案” 根据某种评分方法得出的分数 2️⃣前三部分相当直观 ——…
AIGC 这次浪潮,本质上还是技术驱动的大浪潮 但是跟几家公司的算法负责人交流的时候,能感觉到他们在这次技术浪潮中并没有太多的话语权。 一个原因可能是他们醉心于技术的研究,而无暇顾及大的趋势和竞争的格局。 但如果自己不去顾及这些,就相当于把决策权交给了PM,让PM决定一个技术团队的上限 emmm
To better augment LLMs with context, it makes a lot of sense to organize context not just as a flat list of text chunks, but as a hierarchy of high-level to low-level details. RAPTOR is a super simple but neat idea towards this direction. Hierarchically cluster and summarize the…
To better augment LLMs with context, it makes a lot of sense to organize context not just as a flat list of text chunks, but as a hierarchy of high-level to low-level details. RAPTOR is a super simple but neat idea towards this direction. Hierarchically cluster and summarize the… https://t.co/LFRgYGLCfu
a16z是著名风险投资公司,前些天看orange分享他们的AI投资理念和已投项目PPT,值得仔细阅读理解。 gamma.app/public/a16z-Co… 他们预测AI技术再过10年发展,一定会诞生很多引领行业潮流的新公司,且大概率是toC产品。 新科技将带来平权,让今天的奢侈品变成明天的日常用品。…
视频模型的 “ControlNet” ?视频可控领域卷起来了! 字节发布了 Boximator 的论文,可作为现有视频扩散模型的插件,与 ControlNet 冻结原始权重+仅训练控制模块 的思路一致 通过框选对象 + 定义位置、形状或路径 实现可控生成 有了Boximator,再加上基于 SVD 的 DragNUWA,这下 SVD 的生态要崛起了
视频模型的 “ControlNet” ?视频可控领域卷起来了! 字节发布了 Boximator 的论文,可作为现有视频扩散模型的插件,与 ControlNet 冻结原始权重+仅训练控制模块 的思路一致 通过框选对象 + 定义位置、形状或路径 实现可控生成 有了Boximator,再加上基于 SVD 的 DragNUWA,这下 SVD 的生态要崛起了 https://t.co/bnzVqDYWok
🔗 x.com/i/status/17527… 在 DragNUWA 的 GitHub 主页上,我们发现了即将发布的 1.6 版本模型的一些新特性和效果预览。 看起来,这次的效果和可控性将迈上一个新的台阶!真是太令人期待了!1⃣ 保持人脸特征的一致性 2⃣ 物体和镜头同时运动 3⃣ 非常规拖拽 图片来源:github.com/ProjectNUWA/Dr…
🔗 x.com/i/status/17527… 在 DragNUWA 的 GitHub 主页上,我们发现了即将发布的 1.6 版本模型的一些新特性和效果预览。 看起来,这次的效果和可控性将迈上一个新的台阶!真是太令人期待了!1⃣ 保持人脸特征的一致性 2⃣ 物体和镜头同时运动 3⃣ 非常规拖拽 图片来源:github.com/ProjectNUWA/Dr…
这个项目好,将 LCM 用在视频生成,只需要 4 步推理就可以生成视频。期待放出代码和权重。 从演示来看视频效果也很不错,支持现有 SD 生态 Animatediff 的所有控制方式。 详细介绍: 受到一致性模型(Consistency Model,…
非盈利机构 AllenAI 发布了真正完全开源 LLM “OLMo”,不止模型权重,还包含完整的训练代码、数据集和训练过程,而此前不论是 LLama 或 Mistral 都只公布部分细节。OLMo 为了打破 Nvidia GPU 的垄断,特地在 AMD 和 NVDA GPU 上都训练了一次,证明 LLM 训练是可以用 AMD 的。allenai.org/olmo/olmo-pape…
Parakeet-TDT:超越Whisper的语音识别模型 英伟达和SunoAI研发的模型,是历史版本的进化版,官方宣称目前开源最佳。可商用。 在线体验:huggingface.co/spaces/nvidia/… 模型地址:huggingface.co/nvidia/parakee… 官方博客:nvidia.github.io/NeMo/blogs/202…
🚀We are thrilled to release LLaVA-1.6, with improved reasoning, OCR, and world knowledge. It supports higher-res inputs, more tasks, and exceeds Gemini Pro on several benchmarks! 🤯 It maintains the data efficiency of LLaVA-1.5, and LLaVA-1.6-34B is trained ~1 day with 32 A100s.…
Project textures in UnrealEngine with ComfyUI 🤯 github.com/AlexanderDzhog…
Both the article and code are now open source. Paper: arxiv.org/abs/2401.11708 Code: github.com/YangLing0818/R… Authored by @LingYang_PKU, Zhaochen Yu, @chenlin_meng, @MinkaiX, @StefanoErmon, Bin Cui
Excited to see the growing recognition of LLM flow engineering! Indeed, our ACL'23F paper demonstrates how a carefully engineered LLM flow can surpass the previous SOTA in long-term conversational models, all without additional training! x.com/kangwook_lee/s…
Excited to see the growing recognition of LLM flow engineering! Indeed, our ACL'23F paper demonstrates how a carefully engineered LLM flow can surpass the previous SOTA in long-term conversational models, all without additional training! x.com/kangwook_lee/s… https://t.co/vvbCSbcCwz
SGLang,lmsys的新推理框架。 后端主要是引入了新的KVCache机制提升速度,前端则引入了类似微软Guidance的机制更好的控制LLM输出。
SGLang,lmsys的新推理框架。 后端主要是引入了新的KVCache机制提升速度,前端则引入了类似微软Guidance的机制更好的控制LLM输出。 https://t.co/X2uxcKzBZx
在 ComfyUI 中使用 GragNUWA 复刻 Runway Multi Motion Brush,效果基本上没有差别,而且还更加灵活,可以增加更多细节的运动路径。 GragNUWA 的潜力无限!#ComfyUI #GragNUWA #RunwayMultiMotionBrush x.com/i/status/17490…
在 ComfyUI 中使用 GragNUWA 复刻 Runway Multi Motion Brush,效果基本上没有差别,而且还更加灵活,可以增加更多细节的运动路径。 GragNUWA 的潜力无限!#ComfyUI #GragNUWA #RunwayMultiMotionBrush x.com/i/status/17490…
是一种新的不改变模型权重的微调方法——代理调整(proxy-tuning) 作者解释的很专业,但看起来还是挺复杂的,超出了我的知识范围,这里仅对作者的原文进行翻译: 刘等人提出了一种全新的大语言模型(LLM)微调方法,不需要改变模型权重,这种方法被称为代理调整(proxy-tuning)(参见 Liu et al.…
是一种新的不改变模型权重的微调方法——代理调整(proxy-tuning) 作者解释的很专业,但看起来还是挺复杂的,超出了我的知识范围,这里仅对作者的原文进行翻译: 刘等人提出了一种全新的大语言模型(LLM)微调方法,不需要改变模型权重,这种方法被称为代理调整(proxy-tuning)(参见 Liu et al.…
发现一个画图的好工具,我认为是画图界的 Notion,操作起来也倍儿方便 :whimsical.com
RAG 并没有大家想象得那么复杂,要对技术祛魅,本质上就是 3 个部分组成。 感谢 @fuxiangPro 把之前写的两篇关于 devv.ai RAG 原理的文章整理了一下,这样图非常清晰!
RAG 并没有大家想象得那么复杂,要对技术祛魅,本质上就是 3 个部分组成。 感谢 @fuxiangPro 把之前写的两篇关于 devv.ai RAG 原理的文章整理了一下,这样图非常清晰! https://t.co/SgjFOmO32b
Very happy to announce that VeRA is accepted at @iclr_conf with scores 8,8,8,5! VeRA makes LoRA ~10x more parameter efficient while retaining the same performance & also works for vision! Paper: arxiv.org/abs/2310.11454 Our very light-weight webpage😏: dkopi.github.io/vera/
Very happy to announce that VeRA is accepted at @iclr_conf with scores 8,8,8,5! VeRA makes LoRA ~10x more parameter efficient while retaining the same performance & also works for vision! Paper: arxiv.org/abs/2310.11454 Our very light-weight webpage😏: dkopi.github.io/vera/ https://t.co/X5bCaEdF9E
Jack Riley @jackrileyau
325 Followers 636 Following Rocket Builder. AI, RevOps and Automation Integrator 🧠Rinetoash @rinetoash58854
0 Followers 106 Following Life itself is a journey, we are all worthy and should strive to travel to different lives.Leo @withLeoAI
764 Followers 3K Following AI Superpowers for Teachers The all-in-one AI toolkit for educators. Grade assignments, give feedback, and create teaching material in minutes, not hours.paco xu @xu_paco
963 Followers 1K Following A husband, a father, a Kubernetes contributor, a big football/Valencia fan, a PUBG fan.Elza @ekdorenbosch
985 Followers 1K Following Dream/Adobe/MJ/Basedlabs | Video- editing | Challenges, QT's, Picks & Re-imaginations | Safe Backroom Dev | Odyssey against betrayel 💫|Israa Ali @IsraaAli2077
2K Followers 2K Following Passionate pharmacist, DALLE-3 explorer and technology obsessed.Qinghe Wang @HuaqiangLiu666
35 Followers 59 Following Ph.D Candidate, Dalian University of Technology.Gsdata @Gsdata5566
2K Followers 2K Following AI art, AI technology, AI music… I’m interested in everything about AI. Follow me to show you all the possibilities of AImeng shao @shao__meng
2K Followers 1K Following Developer | Exploring Gen AI 👨💻 Passionate about LLM and T2I 🧠 Share images generated by 👇🏻 Freepik, Ideogram, Stylar and othersやさいはうす @yasaihouse
83 Followers 151 Following AIによるイラスト(bing image Creator)、会話(Claude3)に救いを求め可能性を探りつつ小説をかきます。V系の歌詞で想像した世界へうちの子を落とします。あとnoteも嗜む程度に。透明な言葉の花束を探して……Cecilia Garcia @CeciGarcia000
863 Followers 2K Following AI enthusiast. Innovative resort management services. Mom, grandmother, friend, sister. Books and screenwriting. 🌠Ryan Boyle @_RyanBoyle_
1K Followers 5K Following Tech Enthusiast 👨🏼💻 Aspiring ML Engineer. Frequent Traveler 🌎 Based in Philly & LA, Soon → SF 🌉Harsh Maheshwari @HarshMheshwari
1K Followers 1K Following Enthusiastic about #GenerativeAI #DataScience 🤖 | Constantly curious learner 🌱 | Applied scientist 2 at @amazon | Writer at @medium | @IITKGP GraduateJiawei Liu @Jia_Wei_LIU
219 Followers 504 Following PhD student at ShowLab NUS, working on 3D/4D/video reconstruction, editing, and generation.继小鹏 @Huanghanzhilian
688 Followers 494 Following 🛠️全栈👨💻创业者✨AI⛵️|开源项目: c-shopping WEB全栈 、APP、快应用小程序 https://t.co/W0xF00H3zS、https://t.co/MYnWHRov6M|博客: https://t.co/Z0Mn6XGEu7Yisol Choi @cpis9898
31 Followers 9 FollowingAwesomeYang @youngquantongxu
49 Followers 146 Following 😎 没想好名字公司 CEO/🎭全干工程师/✍️墨问付费专栏作者/著有:🏆插件 https://t.co/FL1h2RHEku /🛠️工具 https://t.co/skjAPmaAOM /🚀AI 导航https://t.co/DLQlAjKGJt 等众多公益项目,想赚美金还没赚到。Vasilije @tricalt
758 Followers 656 Following https://t.co/hFbPLnHIHV | Big Data | Vizsla | Pizza oven | BSc in Psychology (ongoing) and Business (2013)สุรีรัต.. @WKEwN3w8fMs82GY
56 Followers 1K Following เราเจอชะตากรรมแบบไหน ชอบติดตามไว้ก่อนได้นะครับ ผมจะส่งข้อมูลติดต่อไปที่หน้าแรกเป็นระยะๆครับRun-Ze Fan @Vfrz525_
356 Followers 639 Following Research Assistant@GAIR Lab @sjtu1896. NLP/LLMs/Alignment/Instruction Tuning. Looking for a Ph.D. in the 2025 fall (US)Jen Jackson @JenJackson84554
13 Followers 641 FollowingXinyue Wei @SarahWeii
147 Followers 130 Following PhD student at UCSD @HaoSuLabUCSD | Previous intern @AdobeResearchSabrina @sabrinaellingh5
256 Followers 3K FollowingTianyi Zhang @tianyiz2022
39 Followers 109 Following Deep sea autonomous vehicle and machine vision @CarnegieMellon | @TJU1895 @UMich alumniAsimov Meetup @AsimovMeetup
108 Followers 180 Following Asimov Meetup – the underground grassroots of the AI society in Dubai. #AsimovAIMeetupVincentLepetit @VincentLepetit2
179 Followers 201 FollowingLourdes Asuncion @LourdesAsu83677
111 Followers 3K FollowingApril Davis @AprilDavis3098
86 Followers 3K FollowingTerri Thomas @TerriThoma10342
15 Followers 632 FollowingHyeonbin Hwang @ronalhwang
144 Followers 201 Following M.S. Student @kaist_ai https://t.co/bQW6mlGzDNtakiyu @takiyu1025_txu
277 Followers 310 Following 社会人 情報系 CG,CV,NN C++,Python,JavaScript ArchLinux,Xmonad,NeoVim派Versun @VersunPan
473 Followers 777 Following INTJ | 伪全栈 | 在前后端反复横跳 | 运维小能手 | InfoSec爱好者 | Pentest菜鸡 | 54321周刊: https://t.co/E8wXS5CKKF | RSS翻译器: https://t.co/f23UClI6bFGenerative AI @generativeaihub
7K Followers 6K Following Inspired by Algorithms, Powered by Imagination: Unleashing the Potential of Generative AI. #GenerativeAI #deeplearning #AI #MachineLearningNextify @nextify2024
6K Followers 555 Following 某不知名香港创业公司 CEO ,关不关注我的代码 https://t.co/h2isoGsGIV ,我们都是好朋友 🫡Pablo Ruiz Ponce @PabloRuizPonce
42 Followers 139 Following 👨🏻🔬 PhD student in ML & CV 🤖 MSc in Artificial Intelligence 👨🏻🎓 Computer ScientistStylez Morales @Koba_1975
888 Followers 2K Following Making the transition from music into the 3D digital art and VFX arena. Don't mind the mess around here, I'm renovating... ayyyyyyAIformedicine @ai4medicine4
96 Followers 1K FollowingShengqu Cai @prime_cai
467 Followers 267 Following CS PhD student @Stanford; former Research intern @Adobe, MS Computer Science @ETH.Ham Huang @Huang_Ham
230 Followers 425 Following PhD student @Princeton Psych under Drs. Natalia Vélez & Tom Griffiths, studying the computational cognition of human aggregate minds. Before @Penn @CalSierra @sierrambonilla
35 Followers 79 Following phd candidate at @ai_ucl | novel view synthesis/3d reconstruction for surgical scenes @weiss_ucl | seattle - londonBhagyashree Puranik @BhagyashreePu13
41 Followers 244 Following PhD Candidate at UC Santa Barbara, working on robust and fair MLZehuan-Huang @huanngzh
19 Followers 48 Following Master student @ BUAA. Passionate about AIGC and 3D vision.David Park @Davidjpark96
31K Followers 306 Following ceo @ jenni ai (3m users) tweets about growth and my startup learnings!Jakub Tomczak @jmtomczak
8K Followers 999 Following Associate prof @TUeindhoven | Advisor @natinlab1 | Founder Amsterdam AI Sols | Before: @VUamsterdam @UvA_Amsterdam (@wellingmax) @Qualcomm | Opinions my ownClifton Poth @clifapt
371 Followers 312 Following ML Engineer @Cohere | Open source @AdapterHub | prev. @TUDarmstadt @UKPLab Mastodon: https://t.co/m2DKcF7zIHNeil Houlsby @neilhoulsby
4K Followers 318 Following Professional AI researcher; amateur athlete. Senior Staff RS in the Google Deepmind, Zürich. Attempts triathlons.AdapterHub @AdapterHub
1K Followers 1K Following A central repository for pre-trained adapter modules in transformers! Active maintainers: @clifapt @h_sterz @LeonEnglaender @timo_imhof @PfeiffJoMark Huang @markatgradient
466 Followers 139 Following @Gradient_AI_ Democratizing Large Models. Former Quant. Waiting for AGI. https://t.co/ZC0c6oBk3SGradient @Gradient_AI_
2K Followers 42 Following Accelerate AI transformation with Gradient AI Foundry, the most comprehensive solution to deploy autonomous assistants.Yichen (Zach) Wang �.. @YichenZW
168 Followers 181 Following 24Fall Incoming NLP Ph.D. @UChicago @UChicagoCI | Interning @UWNLP @Tsvetshop (@BerkeleyNLP before) | Senior @XJTU1896 Honored CS 24’Elza @ekdorenbosch
985 Followers 1K Following Dream/Adobe/MJ/Basedlabs | Video- editing | Challenges, QT's, Picks & Re-imaginations | Safe Backroom Dev | Odyssey against betrayel 💫|Berrak Sisman @berraksismann
1K Followers 895 Following Assistant Professor, Electrical and Computer Engineering, University of Texas at Dallas @ut_dallas | IEEE Speech and Language Processing Technical CommitteeSavage Lab @SavageCatsOnly
2K Followers 478 Following Compartmentalizing your metabolism, and y’know, CRISPR stuff. Account managed by grad students and postdocs 🤙 @UCBerkeley @berkeleyMCB @igisci @HHMINEWSBoyang "Albert" Li @AlbertBoyangLi
884 Followers 375 Following Nanyang Associate Prof, NRF Fellow, #NTUsg. #AI, #ML, Multimodal, Narrative Intelligence. Formerly Baidu & Disney Research. PhD Georgia Tech.Poppy's Pixels @popped_pixels
510 Followers 452 Following I make things with StableDiffusion, Maya and Adobe software. This page is mostly for AI Art. My 3d and game design stuff is elsewhere.MR.WAS @Shane__Willett
854 Followers 99 Following Art Director and Prompt specialist and Generative Artist and animator. I went to SAIC and have a BFA in New Media.paco xu @xu_paco
963 Followers 1K Following A husband, a father, a Kubernetes contributor, a big football/Valencia fan, a PUBG fan.OpenLLMLeaders @OpenLLMLeaders
190 Followers 1 Following Track 🤗 Open LLM Leaderboard. Created by https://t.co/ywEwEb4O1G𝗦𝘁𝗲𝘃𝗲 @st7evechou
8K Followers 382 Following 我是 Steve 。SteveWatch 开发者 / YouTuber / 数码爱好者 / Tesla Model Y 车主 / Apple Fans / APP 安利推广者/🈲政治。 TG: https://t.co/nDC0kDkpBn Steve Studio:https://t.co/svpNx4BviBAlexandru Voica 💀 @alexvoica
16K Followers 912 Following Corp affairs @SynthesiaIO | Director @BlackKnight_ltd | Advisor @MBZUAI | Guns N' Roses, 🏀, Philip K. Dick, intelligent machines, peanut butter + maple syrupRichard Kim ↙️ @richardkimphd
5K Followers 5K Following "⚡Harbinger of the Generative AI Paradigm Shift 🔭Unveiling Frontier Advancements & Research 👽Sculpting the Techno-Sapien Era"Sirui Chen @eric_srchen
88 Followers 171 Following PhD in Stanford CS, Prev Undergrad at HKU. Interested in roboticsValerio Zerbi @Valerio_Zerbi
1K Followers 1K Following SNSF Eccellenza Professor @EPFL_en Interested in preclinical functional neuroimaging, brain stimulation and modelling in PsychiatryPolytope Labs @PolytopeLabs
3K Followers 10 Following A blockchain research & development lab at the forefronts of the decentralization revolution. Join us https://t.co/K3APG8sMmuRui Shu @_smileyball
3K Followers 395 Following I draw smileyball https://t.co/VZJD2Av8PY Calculating lower bounds @OpenAIZijian "Jason" Ding @jasonzding
1K Followers 1K Following PhD candidate @hcil_umd | Human-GAI Research | Incoming Intern @IBMResearch Previously @MSFTResearch @Dataminr @DesignLabUCSD @hkust @Cambridge_UniKexin Huang @KexinHuang5
2K Followers 561 Following PhD Student @Stanford CS with @jure; Machine Learning + BiomedicineTatsuya Shirakawa @s_tat1204
1K Followers 792 Following Human / 合同会社nouu代表 機械学習や数理最適化、データサイエンスの周辺領域に生息しています。 アドバイザリーや技術支援もしています。 お仕事の依頼/ご相談はお気軽にどうぞ。Winsu @WinsuPro
173 Followers 111 Following 3D Character Artist Open commission to create 3D VTuber Avatar Commission Status : OPEN (Limited Slot)Qinghe Wang @HuaqiangLiu666
35 Followers 59 Following Ph.D Candidate, Dalian University of Technology.Shreya Gupta @ShreyaByte
193 Followers 299 Following AI Evangelist & Entrepreneurial Mind 🚀 | Sharing the latest AI breakthroughs and business insights. Turn business ideas into profit with #AIMauro Comi @mauro_ai
400 Followers 660 Following PhD student in Machine Learning, 3D Computer Vision, and Robotics. Soft spot for graphics research, Pixar movies, and chocolate || https://t.co/SNZDhbDXzvTales @xztales
232 Followers 276 Following Product design , focus on underlying logic、narrative、Ai、science fiction etc. 推特是非线性笔记工具。0xor0ne @0xor0ne
55K Followers 525 Following | CyberSecurity | Reverse Engineering | C and Rust | Exploit | Linux kernel | PhD | My Tweets, My Opinions :) |Blue Nile 3D @blue_nile_3d
978 Followers 4K Following 3d freelance artist and addon creator using Blender https://t.co/fQBeNlxxGTGeleta @geletavc
922 Followers 783 Following Find me at tech parties in SF. Acceleration. Eccentricity. @Dolby, ex-@Amazon, scout @AlumniVentures. PhD @UCBerkeley/@berkeley_ai + @StanfordDBDS on AI+🧬Scott Hanselman 🌮 @shanselman
329K Followers 11K Following VP of Developer Community @ MSFT - Code, OSS, STEM, Beyoncé, 🏴🇿🇼#T1D, #DevRel YouTube+TikTok listen to the @Hanselminutes inclusive tech podcast!neptune.ai @neptune_ai
7K Followers 906 Following The MLOps stack component for experiment tracking. We tweet about #MLOps best practices & other cool stuff. Read our blog at https://t.co/nOTpkA75fEPrince Canuma @Prince_Canuma
2K Followers 910 Following ML Engineer 👨🏾💻 • MLOps • LLMs • RAG • Speaker • Writer • Ex-@neptune_ai • https://t.co/iZnxoefJBUAdi Simhi @AdiSimhi
75 Followers 75 FollowingKristi Hines @kristileilani
49K Followers 4K Following Content creator, freelance writer, GPT builder, and photographer. DM or 📧 [email protected] (business inquiries only)Topology and adaptive Level-of-Detail (LOD) have been unsolved problems and critical roadblocks towards full-scale studio/enterprise adoption of 3D GenAI. A new era in 3D generative modeling is taking shape, beyond SDS-based NeRFs/Gaussian Splats, SDFs, DMTeT, Flexicubes, etc.…
Productive week for Bytedance on Paper Page🔥 ✨ ID-Aligner: huggingface.co/papers/2404.15… ✨ Hyper-SD: huggingface.co/papers/2404.13… ✨ PuLID: huggingface.co/papers/2404.16… ✨ TextSquare: huggingface.co/papers/2404.12… ✨ Groma: huggingface.co/papers/2404.13… huggingface.co/papers?date=20…
应该是这个工具,你们可以试试🙃 DeepFacelive: 可以在直播过程和视频通话时进行实时换脸的工具 DeepFaceLive 建立在 DeepFaceLab 的基础上,后者为当前领先的面部交换框架,能够产生接近电影质量的面部合成效果,提供高保真的视觉体验。…
顺藤摸瓜找到这个视频的作者了 问了下 一对一教学是3000元 丢个视频你自己学是588元 可以使用训练好的模型,也可以自己用SD生成定制自己独有的虚拟人模型 一个模型一张脸,做好无法更换! 显卡3060起步
🚀CharacterFactory: Sampling Consistent Characters with GANs for Diffusion Models We propose CharacterFactory, a framework that allows sampling new characters with consistent identities in the latent space of GANs for diffusion models. 🔥Project page: qinghew.github.io/CharacterFacto…
完成度最高的 AI 搜索引擎/Perplexity Clone llm-answer-engine 开源,目前 3.3K Star ⭐️ 1. 支持搜索展示图片、视频、地图等内容 2. 使用 Vercel AI SDK、Groq、Mixtral、Langchain、Brave & Serper 等前沿技术构建💥 3. 提供文字和 Youtube 视频教程讲解实现原理👍 github.com/developersdige…
Llama中文社区,Llama3在线体验和微调模型已开放, 实时汇总最新Llama3学习资料,已将所有代码更新适配Llama3, 构建最好的中文Llama大模型,完全开源可商用 github.com/zgimszhd61/Lla…
time for a paper post: A Roadmap to Pluralistic Alignment what happens when you take alignment down to its most granular form? you arrive at individual alignment... aka personalization. but to date, most of the utility found in LLMs has come from a global alignment approach…
About two weeks ago, I became captivated by the quality differences between ComfyUI and @diffuserslib results. It led me down a deep rabbit hole, but in addition to understanding these differences, I developed a gem. Optimized for production and incredibly fast—it takes just a…
PuLID:一个新的ID保持项目。 优势是除了 ID 保持更好之外,还最小化了对原始模型的影响。 PuLID 的一个显著特点是,无论是在身份信息加入前后,图像的背景、光线、布局和风格等元素都保持了高度一致。
IDM-VTON:虚拟试衣技术 能够生成高度真实的虚拟试衣图像,细节更加精细。 IDM-VTON能够捕捉到服装的细节,如纹理、图案和缝线等,这些细节在试衣图像中被准确地再现。 即使是在户外或者背景复杂的照片中,这项技术也能准确地展示衣物试穿效果,保持高质量的图像输出。…
重新打光迈向新阶段!IntrinsicAnything:从图像中高质量恢复物体材质 通过将渲染方程分解为漫反射和镜面反射项,研究把材料先验公式化为漫反射和镜面反射的扩散模型,由此可以更准确地从复杂的真实环境图像中恢复材质信息,避免歧义 还通过从粗到细的训练策略进一步提高了模型性能…
英伟达前几天发布了一个巨牛皮的项目Align Your Steps,可以大幅提高 SD 低推理步数生成图像的效果。 而且不只是针对图像生效,对 SVD 模型也有很好的效果。 ComfyUI 已经原生支持了这个算法,试了一下 10 步的时候效果相当好。 可以用我下面👇的工作流尝试。
大模型顶尖团队的绝佳审美:Cohere Design 最近被 Cohere Command R 系列模型、Cohere Toolkit 深深震撼,在感叹这个团队强大的企业模型能力的同时,也被它产品设计的优雅深深吸引,深深的好奇,Cohere 的审美怎么会这么好这么优雅,让我们接着往下看 👇 Cohere Design 由 Pentagram…
开源推荐 Cohere 发布并开源了 Cohere Toolkit,加速 AI 应用开发 👏 可用于生产的应用程序的开源存储库,可以跨云提供商进行部署,这些应用程序可以跨 AWS、Azure 和 Cohere 平台访问 Cohere 的 Command、Embed 和 Rerank 模型。 只需三步就可以快速完成开发部署: 1. 选择模型提供方 2.…
问:大模型在做 RAG 回复问题的时候,怎么稳定的在内容中带上被引用片段的标号呢 这种效果咋实现的 问: 做 RAG 时,是需要对文档预处理的:对文档分块(例如:按照章节或者段落分),然后每一个分块做 Embedding,然后将 Embedding 后的结果存入向量数据库。…
LangChain:构建与数据对话的聊天机器人3——文档分割 #LangChain构建与数据对话的聊天机器人 这节课主要介绍了LangChain文档分割器的使用。讨论了如何将文档分割成更小的语义相关的块,以便进行后续的处理。…
只需 1-2 步!高效的零样本语音合成:FlashSpeech 1)高效:基于 LCM 构建 + 新对抗性训练方法,显著提高生成效率,并保证高质量的语音输出 2)更自然:引入韵律生成模块,增强语音自然度和韵律多样性 3)高速: 1-2个 采样步骤即可,比其他零样本语音合成系统快约 20倍(同时保持声音质量和相似度)…
FlashSpeech Efficient Zero-Shot Speech Synthesis Recent progress in large-scale zero-shot speech synthesis has been significantly advanced by language models and diffusion models. However, the generation process of both methods is slow and computationally intensive.
#国外爆火emo模型国内上线了 阿里可以让人像照片变成唱歌视频的项目 EMO 终于发布了,体验了一下非常强。 一张简单的照片加上克隆的语音模型,字节就可以定制自己的数字人出镜。 Heygen 之类的产品都需要录制一段相当长的视频,并且算力成本也很高,这个直接是免费的。…
文章对Llama3 不同量化方法评估了性能损失,结论和之前文章基本一致: 1. 8bit 量化是免费午餐,无损失。 2. AWQ 4bit量化对8B模型来说有2%性能损失,对70B模型只有0.05%性能损失。可以说也是免费午餐了。 3. 参数越大的模型,低bit量化损失越低。AWQ 3bit 70B 也只有2.7%性能损失,完全可接受。…
" LLAMA3 still suffers non-negligent degradation in these scenarios, especially in ultra-low bit-width. " Very interesting paper in the Large Language Model space named "How Good Are Low-bit Quantized LLAMA3 Models? An Empirical Study" 📌 This research dives deep into…
Introduce OpenVoice V2 - a Text-to-Speech model that can clone any voice and speak in any language. Developed by MyShell and @MIT_CSAIL researchers. 🌐 Imagine your voice going global in multiple languages. 🔊 OpenVoice V2 breaks the language barrier and redefines voice…
@imxiaohu SD只是生成人脸用作替换,实际主要是实时换脸DeepFace Live Github:github.com/iperov/DeepFac… 成品软件:deepfakevfx.com/downloads/deep…