Extracting Skill-Centric State Abstractions from Value Functions
Google at ICLR 2022
Pix2Seq: A New Language Interface for Object Detection
Hidden Interfaces for Ambient Computing
FormNet: Beyond Sequential Modeling for Form-Based Document Understanding
Learning to Prompt for Continual Learning
Locked-Image Tuning: Adding Language Understanding to Image Models
Simple and Effective Zero-Shot Task-Oriented Dialogue
Lidar-Camera Deep Fusion for Multi-Modal 3D Detection
Large-Scale Matrix Factorization on TPUs
VDTTS: Visually-Driven Text-To-Speech
Efficiently Initializing Reinforcement Learning With Prior Policies
Reproducibility in Deep Learning and Smooth Activations
Pathways Language Model (PaLM): Scaling to 540 Billion Parameters for Breakthrough Performance
Introducing CVSS: A Massively Multilingual Speech-to-Speech Translation Corpus
