InfoQ Homepage Google DeepMind Content on InfoQ
-
Google's JEST Algorithm Automates AI Training Dataset Curation and Reduces Training Compute
Google DeepMind recently published a new algorithm for curating AI training datasets: multimodal contrastive learning with joint example selection (JEST), which uses a pre-trained model to score the learnability of batches of data. Google's experiments show that image-text models trained with JEST-curated data require 10x less computation than baseline methods.
-
Recap of Google I/O 2024: Gemini 1.5, Project Astra, AI-powered Search Engine
Google recently hosted its annual developer conference, Google I/O 2024, where numerous announcements were made regarding Google’s apps and services. As anticipated, AI was a focal point of the event, being incorporated into almost all Google products. Here is a summary of the major announcements from the event.
-
Meta and IBM Lead Formation of AI Alliance to Drive Open-Source Innovation
A new consortium, led by Meta and IBM, has been formed to support open-source AI. The AI Alliance comprises numerous organizations from various sectors, including software, hardware, nonprofit, public, and academic, with the goal of collaboratively developing tools and programs to facilitate open development.
-
Google Announces New DeepMind Model, Lyria, to Generate High-Quality Music
Google has introduced Google DeepMind's Lyria, an AI music generation model able to generate vocals, lyrics, and background tracks mimicking the style of popular artists. The model is experimentally available on YouTube through two distinct AI experiments.
-
Google DeepMind Announces LLM-Based Robot Controller RT-2
Google DeepMind recently announced Robotics Transformer 2 (RT-2), a vision-language-action (VLA) AI model for controlling robots. RT-2 uses a fine-tuned LLM to output motion control commands. It can perform tasks not explicitly included in its training data and improves on baseline models by up to 3x on emergent skill evaluations.
-
Google Announces State-of-the-Art PaLM 2 Language Model Powering Bard
Google DeepMind recently announced PaLM 2, a large language model (LLM) powering Bard and over 25 other product features. PaLM 2 significantly outperforms the previous version of PaLM on a wide range of benchmarks, while being smaller and cheaper to run.
-
DeepMind Open-Sources AI Interpretability Research Tool Tracr
Researchers at DeepMind have open-sourced TRAnsformer Compiler for RASP (Tracr), a compiler that translates programs into neural network models. Tracr is intended for research in mechanistic interpretability of Transformer AI models such as GPT-3.
-
DeepMind Announces Minecraft-Playing AI DreamerV3
Researchers from DeepMind and the University of Toronto announced DreamerV3, a reinforcement-learning (RL) algorithm for training AI models for many different domains. Using a single set of hyperparameters, DreamerV3 outperforms other methods on several benchmarks and can train an AI to collect diamonds in Minecraft without human instruction.
-
DeepMind Trains 80 Billion Parameter AI Vision-Language Model Flamingo
DeepMind recently trained Flamingo, an 80B parameter vision-language model (VLM) AI. Flamingo combines separately pre-trained vision and language models and outperforms all other few-shot learning models on 16 vision-language benchmarks. Flamingo can also chat with users, answering questions about input images and videos.
-
DeepMind Introduces Gato, a New Generalist AI Agent
Gato, as the agent is known, is DeepMinds’s generalist AI that can perform many different tasks that humans can do, without carving a niche for itself as an expert on one task. Gato can perform more than 600 different tasks, such as playing video games, captioning images and moving real-world robotic arms. Gato is a multi-modal, multi-task, multi-embodiment generalist policy.
-
DeepMind Trains AI Controller for Nuclear Fusion Research Device
Researchers at Google subsidiary DeepMind and the Swiss Plasma Center at EPFL have developed a deep reinforcement learning (RL) AI that creates control algorithms for tokamak devices used in nuclear fusion research. The system learned control policies while interacting with a simulator, and when used to control a real device was able to achieve novel plasma configurations.
-
DeepMind Open-Sources Quantum Chemistry AI Model DM21
Researchers at Google subsidiary DeepMind have open-sourced DM21, a neural network model for mapping electron density to chemical interaction energy, a key component of quantum mechanical simulation. DM21 outperforms traditional models on several benchmarks and is available as an extension to the PySCF simulation framework.
-
Google Trains 280 Billion Parameter AI Language Model Gopher
Google subsidiary DeepMind announced Gopher, a 280-billion-parameter AI natural language processing (NLP) model. Based on the Transformer architecture and trained on a 10.5TB corpus called MassiveText, Gopher outperformed the current state-of-the-art on 100 of 124 evaluation tasks.
-
DeepMind Releases Weather Forecasting AI Deep Generative Models of Rainfall
DeepMind open-sourced a dataset and trained model snapshot for Deep Generative Models of Rainfall (DGMR), an AI system for short-term precipitation forecasts. In evaluations conducted by 58 expert meteorologists comparing it to other existing methods, DGMR was ranked first in accuracy and usefulness in 89% of test cases.
-
DeepMind Open Sources Data Agnostic Deep Learning Model Perceiver IO
DeepMind has open-sourced Perceiver IO, a general-purpose deep-learning model architecture that can handle many different types of inputs and outputs. Perceiver IO can serve as a "drop-in" replacement for Transformers that performs as well or better than baseline models, but without domain-specific assumptions.