Gato
A multi-task embodied agent model developed by Google DeepMind that unifies a broad range of tasks—including visual-language understanding and robotic arm control—within a single generalist neural network architecture.
Core Facts
Timeline (last 90 days)
Google DeepMind's Gato model unifies image patches, text tokens, and discretized actions into the same sequence.
Google DeepMind's Gato model demonstrated the possibility of unifying image patches, text tokens, and discretized actions into a single sequence.
Google's RT-2 and DeepMind's Gato projects are exploring vision-language-action models.
All Facts (4)
Institutions such as DeepMind and Meta AI have produced representative world model works including DreamerV3 and JEPA
65%UnverifiedGoogle DeepMind's Gato model unifies image patches, text tokens, and discretized actions into the same sequence.
50%UnverifiedGoogle DeepMind's Gato model demonstrated the possibility of unifying image patches, text tokens, and discretized actions into a single sequence.
50%UnverifiedGoogle's RT-2 and DeepMind's Gato projects are exploring vision-language-action models.
50%