Abstract: Systems and methods for flexible parameter sharing for multi-task learning are provided. A training method can include obtaining a test input, selecting a particular task from one or more tasks, and training a multi-task machine-learned model for the particular task by performing a forward pass using the test input and one or more connection probability matrices to generate a sample distribution of test outputs, training the components of the machine-learned model based at least in part on the sample distribution, and performing a backwards pass to train a connection probability matrix of the multi-task machine-learned model using a straight-through Gumbel-softmax approximation.

Abstract: A system includes a neural network that includes a Mixture of Experts (MoE) subnetwork between a first neural network layer and a second neural network layer. The MoE subnetwork includes multiple expert neural networks. Each expert neural network is configured to process a first layer output generated by the first neural network layer to generate a respective expert output. The MoE subnetwork further includes a gating subsystem that selects, based on the first layer output, one or more of the expert neural networks and determine a respective weight for each selected expert neural network, provides the first layer output as input to each of the selected expert neural networks, combines the expert outputs generated by the selected expert neural networks in accordance with the weights for the selected expert neural networks to generate an MoE output, and provides the MoE output as input to the second neural network layer.

Abstract: A system includes a neural network that includes a Mixture of Experts (MoE) subnetwork between a first neural network layer and a second neural network layer. The MoE subnetwork includes multiple expert neural networks. Each expert neural network is configured to process a first layer output generated by the first neural network layer to generate a respective expert output. The MoE subnetwork further includes a gating subsystem that selects, based on the first layer output, one or more of the expert neural networks and determine a respective weight for each selected expert neural network, provides the first layer output as input to each of the selected expert neural networks, combines the expert outputs generated by the selected expert neural networks in accordance with the weights for the selected expert neural networks to generate an MoE output, and provides the MoE output as input to the second neural network layer.

Abstract: Systems and methods for flexible parameter sharing for multi-task learning are provided. A training method can include obtaining a test input, selecting a particular task from one or more tasks, and training a multi-task machine-learned model for the particular task by performing a forward pass using the test input and one or more connection probability matrices to generate a sample distribution of test outputs, training the components of the machine-learned model based at least in part on the sample distribution, and performing a backwards pass to train a connection probability matrix of the multi-task machine-learned model using a straight-through Gumbel-softmax approximation.

Abstract: A system includes a neural network that includes a Mixture of Experts (MoE) subnetwork between a first neural network layer and a second neural network layer. The MoE subnetwork includes multiple expert neural networks. Each expert neural network is configured to process a first layer output generated by the first neural network layer to generate a respective expert output. The MoE subnetwork further includes a gating subsystem that selects, based on the first layer output, one or more of the expert neural networks and determine a respective weight for each selected expert neural network, provides the first layer output as input to each of the selected expert neural networks, combines the expert outputs generated by the selected expert neural networks in accordance with the weights for the selected expert neural networks to generate an MoE output, and provides the MoE output as input to the second neural network layer.

Abstract: A system includes a neural network that includes a Mixture of Experts (MoE) subnetwork between a first neural network layer and a second neural network layer. The MoE subnetwork includes multiple expert neural networks. Each expert neural network is configured to process a first layer output generated by the first neural network layer to generate a respective expert output. The MoE subnetwork further includes a gating subsystem that selects, based on the first layer output, one or more of the expert neural networks and determine a respective weight for each selected expert neural network, provides the first layer output as input to each of the selected expert neural networks, combines the expert outputs generated by the selected expert neural networks in accordance with the weights for the selected expert neural networks to generate an MoE output, and provides the MoE output as input to the second neural network layer.

Abstract: A system includes a neural network that includes a Mixture of Experts (MoE) subnetwork between a first neural network layer and a second neural network layer. The MoE subnetwork includes multiple expert neural networks. Each expert neural network is configured to process a first layer output generated by the first neural network layer to generate a respective expert output. The MoE subnetwork further includes a gating subsystem that selects, based on the first layer output, one or more of the expert neural networks and determine a respective weight for each selected expert neural network, provides the first layer output as input to each of the selected expert neural networks, combines the expert outputs generated by the selected expert neural networks in accordance with the weights for the selected expert neural networks to generate an MoE output, and provides the MoE output as input to the second neural network layer.