Google Inc.

Osaro, Inc.

Ben David Goodrich

Noam M Shazeer

Mohammad Saleh

Etienne Pot

Lukasz Mieczyslaw Kaiser

Peter Junteng Liu

Ryan Sepassi

Vishaal Prasad

Chinnadhurai Sankar

Daniel Duckworth

Arvind Neelakantan

Semih Yavuz

William Lance Richards

Rishi Sharma

Alex Kuefler

Christopher Correa

Sulabh Kumra

Kumar Chippala

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating an output sequence from an input sequence. One of the methods includes, at each of a plurality of generation time steps: generating a combined sequence for the generation time step that includes the input sequence followed by the output tokens that have already been generated as of the generation time step; processing the combined sequence using a self-attention decoder neural network to generate a time step output that defines a score distribution over a set of possible output tokens; and selecting, using the time step output, an output token from the set of possible output tokens as the next output token in the output sequence.

Attention-based decoder-only sequence transduction neural networks

A computer system trains a neural network to predict, for each pixel in an input image, the position that a robot's end effector would reach if a grasp ('poke') were attempted at that position. Training data consists of images and end effector positions recorded while a robot attempts grasps in a pick-and-place environment. For an automated grasping policy, the approach is self-supervised, as end effector position labels may be recovered through forward kinematics, without human annotation. Although gathering such physical interaction data is expensive, it is necessary for training and routine operation of state of the art manipulation systems. Therefore, the system comes “for free” while collecting data for other tasks (e.g., grasping, pushing, placing). The system achieves significantly lower root mean squared error than traditional structured light sensors and other self-supervised deep learning methods on difficult, industry-scale jumbled bin datasets.

Computer-automated robot grasp depth estimation

Methods, systems, and media for determining playlist title coherence and quality are provided. In some embodiments, a method for generating playlist recommendations includes: determining, using a hardware processor, a title of a playlist; generating, using the hardware processor, a byte-level representation of the title based on the title of the playlist; determining, using the hardware processor, an embedded representation of the title based on the byte-level representation; determining, using the hardware processor, a perplexity score of the title by inputting the embedded representation of the title into a trained language model, wherein the perplexity score is an output of the trained language model; and causing, using the hardware processor, a recommendation based on the perplexity score of the title to be presented.

Methods, systems, and media for determining playlist title coherence and quality

Training and/or utilizing a single neural network model to generate, at each of a plurality of assistant turns of a dialog session between a user and an automated assistant, a corresponding automated assistant natural language response and/or a corresponding automated assistant action. For example, at a given assistant turn of a dialog session, both a corresponding natural language response and a corresponding action can be generated jointly and based directly on output generated using the single neural network model. The corresponding response and/or corresponding action can be generated based on processing, using the neural network model, dialog history and a plurality of discrete resources. For example, the neural network model can be used to generate a response and/or action on a token-by-token basis.

Generating automated assistant responses and/or actions directly from dialog history and resources

A computer system trains a neural network on an instance segmentation task by casting the problem as one of mapping each pixel to a probability distribution over arbitrary instance labels. This simplifies both the training and inference problems, because the formulation is end-to-end trainable and requires no post-processing to extract maximum a posteriori estimates of the instance labels.

Instance segmentation by instance label factorization

Growing community of inventors

San Francisco, CA, United States of America

Ben David Goodrich