AIGC周刊 | 第16期

时间： 2023.2.20-2023.2.26

最新技术：

• Modulating Pretrained Diffusion Models for Multimodal Image Synthesis

  论文: https://arxiv.org/abs/2302.12764

  主页: https://mcm-diffusion.github.io

• Check Your Facts and Try Again: Improving Large Language Models with External Knowledge and Automated Feedback

  论文: https://arxiv.org/abs/2302.12813

• DiffusioNeRF: Regularizing Neural Radiance Fields with Denoising Diffusion Models

  论文: https://arxiv.org/abs/2302.12231

• More than you've asked for: A Comprehensive Analysis of Novel Prompt Injection Threats to Application-Integrated Large Language Models

  论文：https://arxiv.org/abs/2302.12173

• pix2pix-zero: Zero-shot Image-to-Image Translation

  主页：https://pix2pixzero.github.io/

• 代码：https://github.com/pix2pixzero/pix2pix-zero

  demo：https://huggingface.co/spaces/pix2pix-zero-library/pix2pix-zero-demo

• ChatGPT for Robotics: Design Principles and Model Abilities

  论文： https://www.microsoft.com/en-us/research/uploads/prod/2023/02/ChatGPT___Robotics.pdf

  报告：https://www.microsoft.com/en-us/research/group/autonomous-systems-group-robotics/articles/chatgpt-for-robotics/

• RealFusion 360◦ Reconstruction of Any Object from a Single Image

  论文: https://arxiv.org/abs/2302.10663

• Zero-Shot Information Extraction via Chatting with ChatGPT

  论文: https://arxiv.org/abs/2302.10205

• Optical Transformers

  论文: https://arxiv.org/abs/2302.10360

• Offline Reinforcement Learning for Mixture-of-Expert Dialogue Management

  论文: https://arxiv.org/abs/2302.10850

• 阿里50亿参数AI画画模型火了！将图像拆分再自由重组 https://mp.weixin.qq.com/s/o93hpXBg63j7jfL279MSlw

• 谷歌复用30年前经典算法，CV引入强化学习，网友：视觉RLHF要来了？ https://mp.weixin.qq.com/s/SE7Lt_o33ofVYEXczkaaaw

• OpenAI发布通用人工智能路线图：AGI比想象中来得更快 https://mp.weixin.qq.com/s/zu1a9p3nDTdk_lZ_-y8XFA

• 单卡就能跑的大模型等效GPT-3！Meta发布大语言模型LLaMA，大小从7B到65B不等

  https://mp.weixin.qq.com/s/PEc12gEzIgv2RKxfj57FOA

• 抖音超900万人在用的「卡通脸」特效技术揭秘 https://mp.weixin.qq.com/s/TN7D9_im-4G9-zASAjlv-Q

• 四少女火爆外网！ControlNet组合拳效果惊人，颠覆AI绘画游戏规则 https://mp.weixin.qq.com/s/HpziNAqHY9Oetsgk2AVxEg

• 从ChatGPT看知识管理及知识图谱的发展

  https://mp.weixin.qq.com/s/_KGLlG67a-C7qKb3Gd902A

课程：

• 李宏毅：ChatGPT 原理剖析

  1.对ChatGPT的常见误解

  2.预训练

  3.ChatGPT所带来的研究问题

• 深入了解视觉语言模型

  https://mp.weixin.qq.com/s/DBAo6TxbVb5Tn2VqTn4xZg

商业：

• 揭秘腾讯大模型新项目组「混元助手」人才地图 https://mp.weixin.qq.com/s/rdpGZII3pu3MHr-lFm3GyQ

• 前京东AI掌门人入局ChatGPT！「不设上限」招揽中国OpenAI合伙人 https://mp.weixin.qq.com/s/w0sjUi7MbnAyxPnuLzMyRg

• ChatGPT群雄逐鹿：陆奇屠龙，号令天下；小川不出，谁与争锋 https://mp.weixin.qq.com/s/NNJzd2lScGlRgl4rKcyx3A

• ChatGPT背后：从0到1，OpenAI的创立之路

  https://mp.weixin.qq.com/s/E1_30D9Jw1XHBQnrrSh4NQ

案例：

• 运用AIGC人工智能生产内容

  https://mp.weixin.qq.com/s/b10sqhA9-IvUT44U9-5dLA

• Runway，新的“新媒体”工具 https://mp.weixin.qq.com/s/OawtlYKZvMY2etZv3lIP6g

• AI画的美女已经在小红书上有1.5万粉丝了 https://mp.weixin.qq.com/s/seFxCaPZSiWIcBHPxfnTKg