2020 PyTorch 开发者日

从今年开始，PyTorch 官方计划举办两个单独的活动: 一个是为开发人员和用户讨论核心技术开发、idea 和 roadmap 准备的活动，称为“开发者日（Developer Day）” ; 另一个是为 PyTorch 生态系统和行业社区展示他们的工作并发现合作机会而举办的活动，称为“生态系统日（Ecosystem Day）”(定于2021年初)。

PyTorch 开发者日将于美西时间 2020年11月12日上午8点（北京时间13日0点）开始，因为疫情原因全线上举办，届时将有一整天关于各种主题的技术演讲，包括支持跨多个领域开发的核心框架、新工具和库的更新。你还将看到关于机器学习系统和工具的最新研究的讨论。所有演讲都将进行现场直播，向公众免费开放，有兴趣的请科学访问这个链接：http://fb.me/e/4wgulWHs9

时间表如下：

开幕

8:00–8:05 AM | Welcome — Joe Spisak, Product Manager, Facebook AI

主题演讲

8:05–8:20 AM | Open Challenges in Deep Learning Systems — Kim Hazelwood, West Coast Head of Engineering, Facebook AI

8:20–8:30 AM | State of PyTorch 2020 — Edward Yang, Research Engineer, Facebook AI

8:30–8:45 AM | Vision of PyTorch: Focus of the next 3 years — Lin Qiao, Director of Engineering, Facebook AI

PyTorch 核心深入研究

8:45–8:53 AM | Complex Numbers in PyTorch — Anjali Chourdia, Software Engineer

8:53–9:01 AM | Making PyTorch More “NumPy Compatible” — Mike Ruberry, Software Engineer

9:01–9:10 AM | High-Level API for Autograd — Alban Desmaison, Research Engineer

9:10–9:20 AM | PyTorch Distributed RPC — Shen Li, Research Scientist

9:20–9:30 AM | PyTorch Distributed Data Parallel (DDP) — Pritam Damania, Software Engineer

9:30–9:36 AM | TorchAudio — Vincent Quenneville-Belair, Machine Learning Scientist

9:36–9:43 AM | TorchText — George Zhang, Software Engineer

9:43–9:50 AM | TorchVision — Francisco Massa, Research Engineer

9:50–9:55 AM | Latest Profiler APIs and Best Practices — Ilia Cherniasvskii, Software Engineer

9:55–10:00 AM | Visualization with TensorBoard — Siqi Yan, Software Engineer

10:00–10:10 AM | PyTorch Performance — Natalia Gimelshein, Applied Research Scientist

10:10–10:20 AM | PyTorch on Windows — Maxim Lukiyanov — Principal Product Manager, Microsoft Azure

10:20–10:30 AM | PyTorch Mobile, David Reiss, Software Engineer

10:30–10:40 AM | PyTorch Mobile and Android Neural Networks API — Oli Gaymond, Product Manager, Android Machine Learning

休息 | 10:40–10:45 AM

PyTorch 研究演讲

10:45–11:00 AM |Torch for R & Hasktorch: Bringing Torch to New Programming Languages — Austin Huang, Vice President, AI & Machine Learning at Fidelity and Daniel Falbel, Software Engineer, RStudio

11:00–11:10 AM | Graph Convolutional Operators in the PyTorch JIT — Lindsey Gray, Scientist, FermiLab and Matthias Fey, Ph.D. Student at TU Dortmund

11:10–11:20 AM| Model Interpretability — Narine Kokhlikyan, Researcher Scientist, Facebook AI

11:20–11:30 AM | Hyperparameter Importance — Crissman Loomis, Engineer/Business Dev, Preferred Networks

11:30 AM — 11:40 AM| DeepSpeed, Yuxiong He, Partner Research Manager at Microsoft

11:40–11:50 AM | MLPerf & PyTorch at NVIDIA — Christian Sarofeen, Manager of PyTorch, NVIDIA

11:50–12:00 PM | Inspirations from Going Abroad — Thomas Viehmann, PyTorch Guru, MathInf GmbH

生产生态系统

12:00–12:10 PM | Reproducible AI using PyTorch and MLFlow — Geeta Ghauhan, PyTorch Partner Engineering, Facebook AI

12:10–12:20 PM | PyTorch/XLA Internal — Jack Cao, Software Engineer, Google

12:20–12:30 PM | TorchServe: Model Server for PyTorch — Lokesh Gupta, Software Dev Manager, AWS

隐私 AI

12:30–12:40 PM | Differential Privacy on PyTorch — Davide Testuggine, Applied Scientist, Facebook AI

12:40–12:50 PM | The future of AI Tools — Andrew Trask, Leader of OpenMined and Ph.D. Student at Oxford University

12:50–1:00 PM| Private AI Education Series Announcement — Andrew Trask, Leader of OpenMined and Joe Spisak, PyTorch Product Manager, Facebook AI

1:00–1:05 PM | Closing — Joe Spisak, PyTorch Product Manager, Facebook AI

关注中，有意思的内容我到时会搬运并解读一下。