Google patents image editing method using delta denoising scores

1. A method performed by one or more computers, the method comprising: obtaining training data that comprises a plurality of training examples, wherein each training example corresponds to a respective image editing task from a set of one or more image editing tasks and comprises a respective initial image, a respective text description of the respective initial image, and a respective target text description for the initial image that describes a target image that is generated by performing the corresponding image editing task on the respective initial image; training an image-to-image translation neural network on the training data, wherein the image-to-image translation neural network is configured to receive a network input comprising an input image and to process the network input to generate an output image, and wherein the training comprises, at each of a plurality of training iterations: obtaining, from the training data, a batch of one or more training examples for the training iteration; for each of the one or more training examples: processing a network input comprising the initial image in the training example using the image-to-image translation neural network to generate a target image; generating a denoising output for the initial image, comprising processing a first diffusion input that comprises the initial image and the text description for the initial image using a trained diffusion neural network to generate an initial denoising output for the initial image; generating a denoising output for the target image, comprising processing a second denoising input comprising the target image and the text description for the target image using the trained diffusion neural network to generate an initial denoising output for the target image; and determining, using the denoising outputs for the initial image and the target image, a gradient of a loss function with respect to the parameters of the image-to-image translation neural network; and updating the network parameters of the image-to-image translation neural network using the gradients of the loss function for the one or more initial images.