LaMDA: Towards Safe, Grounded, and High-Quality Dialog Models for Everything
Learning to Route by Task for Efficient Inference
A Fast WordPiece Tokenization System
More Efficient In-Context Learning with GLaM
Evaluating Syntactic Abilities of Language Models
Predicting Text Readability from Scrolling Interactions
GoEmotions: A Dataset for Fine-Grained Emotion Classification
SimVLM: Simple Visual Language Model Pre-training with Weak Supervision
Introducing FLAN: More generalizable Language Models with Instruction Fine-Tuning
The C4_200M Synthetic Dataset for Grammatical Error Correction
From Vision to Language: Semi-supervised Learning in Action…at Scale
