Live caption systems often update predictions as new content is received, which can cause “flickers” of text instability that impair the reading experience. Read how we improve text stability in live captions w/ a new flicker metric & stability algorithm ↓
https://
goo.gle/47VrtKn
@googleai
-
Google Improves Live Caption Stability With New Flicker Metric
By
–
-
Natural Language Commands Control Four-Legged Robot Locomotion
By
–
Learn how a four-legged agent’s foot contact patterns were used as a new interface between natural language user commands and a locomotion controller to build an interactive robot system that allows users to flexibly craft locomotion behaviors → https://t.co/t9vyjdt1f6 pic.twitter.com/sENarsGQKY
— Google AI (@GoogleAI) 29 août 2023Learn how a four-legged agent’s foot contact patterns were used as a new interface between natural language user commands and a locomotion controller to build an interactive robot system that allows users to flexibly craft locomotion behaviors → https://
goo.gle/3EfEXD0 -
RO-ViT: Region-Aware Vision Transformer Pre-Training Method
By
–
Check out RO-ViT, a simple method to pre-train vision transformers in a region-aware manner (using a novel technique called “cropped positional embeddings”) to improve open-vocabulary detection. Learn more and grab the code at https://
goo.gle/3YB7cFC. -
Google’s Perception Fairness Team Advances Multimodal AI Inclusion
By
–
Today we highlight the work of Google’s Perception Fairness team, which works to advance the frontiers of fairness and inclusion in multimodal #MachineLearning systems, especially related to foundation models and #GenerativeAI. Learn about this important work at… pic.twitter.com/tssgeA8Yrq
— Google AI (@GoogleAI) 25 août 2023Today we highlight the work of Google’s Perception Fairness team, which works to advance the frontiers of fairness and inclusion in multimodal #MachineLearning systems, especially related to foundation models and #GenerativeAI. Learn about this important work at
-
Quantum Computing Progress: New Framework for Current Noisy Processors
By
–
Today we discuss our progress in developing useful beyond-classical quantum experiments that can be performed on current noisy quantum processors, including a new framework for measuring the computational cost of a quantum experiment → https://
goo.gle/45o4B4E -
In-Context Learning Enables Algorithmic Reasoning in LLMs
By
–
Today on the blog, learn about an approach that leverages in-context learning and a novel algorithmic prompting technique to enable algorithmic reasoning capabilities (i.e., applying a set of abstract rules to solve a task) in large language models → https://t.co/ldbWs9Cclx pic.twitter.com/r0e6Q73AjK
— Google AI (@GoogleAI) 24 août 2023Today on the blog, learn about an approach that leverages in-context learning and a novel algorithmic prompting technique to enable algorithmic reasoning capabilities (i.e., applying a set of abstract rules to solve a task) in large language models → https://
goo.gle/3qOMmGr -
NeurIPS 2023 ATTRIB Workshop: Understanding ML Model Behavior Attribution
By
–
Recent innovation has given rise to #ML models w/ impressive capabilities, but there’s much to learn about how we attribute model behavior to training data, algorithms, architecture, & more! Have papers or ideas on this? Submit to ATTRIB @ #NeurIPS2023 → http://
attrib-workshop.cc -
TokenSplit Discussion at Interspeech 2023 Google Booth
By
–
Miss the discussion of TokenSplit on Tuesday? You'll have another chance to hear @ScottTWisdom & @HakanErdoganPhD talk about it today. Just drop by the #Interspeech2023 Google booth at 3:30pm to learn all about it.
-
Language-to-Reward System Enables Natural Language Robot Control
By
–
Introducing a new language-to-reward system for interfacing LLMs with robots using reward functions. Learn how the system’s predictive control tool enables users to teach robots novel actions using natural language inputs → https://t.co/cLPgwuvvon pic.twitter.com/fxkIKv8bgR
— Google AI (@GoogleAI) 22 août 2023Introducing a new language-to-reward system for interfacing LLMs with robots using reward functions. Learn how the system’s predictive control tool enables users to teach robots novel actions using natural language inputs → https://
goo.gle/44kjHXx -
TokenSplit: Discrete Token Speech Separation Model at Interspeech
By
–
At 3:30pm today, drop by the #Interspeech2023 Google booth to hear @ScottTWisdom & @HakanErdoganPhD discuss TokenSplit, a speech separation model that uses discrete token sequences representing audio signals & can improve output by using transcripts or separated audio inputs.