Digest 2022-01 # Speech AI that understands speech by looking as well as hearing - ai.facebook.com/blog/ai…-hearing HuBERT: Self-supervised representation learning for speech recognition, generation, and compression - ai.facebook.com/blog/hu…pression # ML Графовые нейронные сети - https://dyakonov.org/2021/12/30/gnn/ A Gentle Introduction to Graph Neural Networks - https://distill.pub/2021/gnn-intro/ GPT-3, Foundation Models, and AI Nationalism - lastweekin.ai/p/gpt-3…ionalism The Illustrated Retrieval Transformer - jalammar.github.io/illustr…nsformer You get what you measure: New NLU benchmarks for few-shot learning and robustness evaluation - www.microsoft.com/en-us/r…aluation Azure AI milestone: New foundation model Florence v1.0 advances state of the art, topping popular computer vision leaderboards - www.microsoft.com/en-us/r…-the-art Language modelling at scale: Gopher, ethical considerations, and retrieval - deepmind.com/blog/ar…at-scale Sequence-to-sequence learning with Transducers - lorenlugosch.github.io/posts/2…ansducer A contemplation of logsumexp - lorenlugosch.github.io/posts/2…ogsumexp Meta claims its AI improves speech recognition quality by reading lips - venturebeat.com/2022/01…ing-lips Training 100B models is fucking hard - github.com/bigscie…arned.md Scaling Vision with Sparse Mixture of Experts - ai.googleblog.com/2022/01…-of.html Интерпретация моделей и диагностика сдвига данных: LIME, SHAP и Shapley Flow - https://habr.com/ru/company/ods/blog/599573/ A ConvNet for the 2020s - https://arxiv.org/pdf/2201.03545.pdf LaMDA: Towards Safe, Grounded, and High-Quality Dialog Models for Everything - ai.googleblog.com/2022/01…igh.html Separating Birdsong in the Wild for Classification - ai.googleblog.com/2022/01…for.html Accurate Alpha Matting for Portrait Mode Selfies on Pixel 6 - ai.googleblog.com/2022/01…ait.html The Gradient Update #16: China's World-leading Surveillance Research and a ConvNet for the 2020s - thegradientpub.substack.com/p/the-g…as-world Does Gradient Flow Over Neural Networks Really Represent Gradient Descent? - http://www.offconvex.org/2022/01/06/gf-gd/ Does Your Medical Image Classifier Know What It Doesn’t Know? - ai.googleblog.com/2022/01…now.html Introducing Text and Code Embeddings in the OpenAI API - openai.com/blog/in…beddings Steering Towards Effective Autonomous Vehicle Policy - thegradient.pub/engagin…gagement Introducing StylEx: A New Approach for Visual Explanation of Classifiers - ai.googleblog.com/2022/01…for.html - https://www.youtube.com/watch?v=mbrka3vBjH8 - tldr very cool, but most likely requires a lot of compute