Abstract: The disclosed method for training a first machine learning model includes generating, based on training data, first output data using a first teacher machine learning model included in one or more teacher machine learning models, generating, based on the training data, second output data using the first machine learning model, wherein the first machine learning model comprises a second machine learning model and one or more low-rank adaptation (LoRA) towers, calculating, based on the first output data and the second output data, a loss, generating, based on the loss, one or more gradients, generating, based on the one or more gradients, one or more LoRA tower ranks, and updating, based on the loss and the one or more LoRA tower ranks, one or more parameters of the one or more LoRA towers.

Abstract: Apparatuses, systems, and techniques to train one or more neural networks using unannotated images. In at least one embodiment, the one or more neural networks are trained based, at least in part, on one or more loss functions calculated using a randomly selected portion pair from two different images and a randomly selected portion pair from the same image.

Abstract: Apparatuses, systems, and techniques to train one or more neural networks using unannotated images. In at least one embodiment, the one or more neural networks are trained based, at least in part, on one or more loss functions calculated using a randomly selected patch pair on a same image and a spatial relationship between two patches within the randomly selected patch pair on the same image.

Abstract: Apparatuses, systems, and methods to use one or more neural networks to generate information about one or more images based, at least in part, on one or more confidence scores associated with the information. In at least one embodiment, a neural network downscales an image and performs an image processing task on said downscaled image according to a query input by a user.

Abstract: The rise of specialized vision foundation models has created a need for methods to consolidate knowledge from multiple models (i.e. the teachers) into a single model (i.e. the student). However, this type of knowledge agglomeration leaves open several critical challenges, including that teacher models typically operate at varying resolutions due to different architectures and training goals, creating feature granularity inconsistencies, that existing models have different distribution moments which can result in biased learning, and that computer vision models are oftentimes trained to produce features at a particular resolution, and therefore do not generalize well to different tasks requiring different resolutions.

Abstract: Apparatuses, systems, and techniques to use one or more neural networks with adjusted resolution information. In at least one embodiment, for example, one or more neural network low resolution encoders are trained to one or more higher resolutions. In at least one embodiment, as another example, a processor is to adjust a resolution of information to be used by one or more neural networks based, at least in part, on one or more performance metrics of one or more neural networks.

Abstract: Apparatuses, systems, and techniques to perform a neural network to generate an image. In at least one embodiment, for example, one or more neural networks generate one or more portions of one or more images and one or more captions. In at least one embodiment, as another example, a processor uses one or more neural networks to generate one or more images from text based, at least in part, on one or more first images without text indicating content of one or more first images and one or more second images with text indicating content of one or more second images.

Abstract: One embodiment of a method for training a first machine learning model includes processing first data via a plurality of trained machine learning models to generate a plurality of first outputs, processing the first data via the first machine learning model to generate a second output, processing the second output via a plurality of projection heads to generate a plurality of third outputs, computing a plurality of losses based on the plurality of first outputs and the plurality of third outputs, and performing one or more operations to update one or more parameters of the first machine learning model and one or more parameters of the plurality of projection heads based on the plurality of losses.

Abstract: Transformers are neural networks that learn context and thus meaning by tracking relationships in sequential data. The main building block of transformers is self-attention which allows for cross interaction among all input sequence tokens with each other. This scheme effectively captures short-and long-range spatial dependencies and imposes time and space quadratic complexity in terms of the input sequence length, which enables their use with Natural Language Processing (NLP) and computer vision tasks. While the training parallelism of transformers allows for competitive performance, unfortunately the inference is slow and expensive due to the computational complexity. The present disclosure provides a computer vision retention model that is configured for both parallel training and recurrent inference, which can enable competitive performance during training and fast and memory-efficient inferences during deployment.

Abstract: Apparatuses, systems, and methods to use one or more neural networks to generate information about one or more images based, at least in part, on one or more confidence scores associated with the information. In at least one embodiment, a neural network downscales an image and performs an image processing task on said downscaled image according to a query input by a user.

Abstract: Apparatuses, systems, and techniques to use one or more neural networks to generate an upsampled version of one or more images based, at least in part, on a denoised version of said one or more images. At least one embodiment pertains to generating an upsampled high-resolution image from a noisy version and denoised version of a low-resolution image. At least one embodiment pertains to separating components of a low-resolution image before denoising an image.

Abstract: Apparatuses, systems, and techniques are presented to reconstruct one or more images. In at least one embodiment, one or more circuits are to use one or more neural networks to adjust one or more pixel blending weights.

Abstract: Methods for composition aware image search refinement using relevance feedback. A method includes receiving, in response to user input, a relevance area within a plurality of image results, wherein the plurality of image results are provided for display on a user interface in response to a first search query using a plurality of spatial anchors in association with respective semantic datasets, and wherein the relevance area is associated with a target semantic dataset. The method also includes determining a mapping from the target semantic dataset to the plurality of spatial anchors, adjusting, based on the mapping, the semantic datasets associated with the plurality of spatial anchors, and providing for display, in the user interface, a plurality of refined image results in response to a second search query using the plurality of spatial anchors in association with the adjusted semantic datasets.

Abstract: A method for receiving a first text in a source language is provided. The method includes associating the first text in the source language with a first vector, selected from a plurality of vectors associated with the source language in an embedded set, and identifying a second vector that is closer to the first vector than a pre-selected threshold. The second vector is associated with a second text in a target language. The method includes associating the first text in the source language with the second text in the target language, wherein the embedded set includes a first image vector for a first image and a second image vector for a second image, and returning the second text in the target language, the second text being a translation of the first text from the source language to the target language. A system configured to perform the above method is also provided.

Abstract: A computer-implemented method is provided for retrieving an image from a user in a desired format and for detecting a compression efficiency for the image. When the compression efficiency is above a pre-selected threshold the computer-implemented method includes obtaining a saliency representation of the image, capturing a feature description of a non-salient portion of the image, flattening the non-salient portion in a new image, storing the new image in a selected format in a memory and storing a background descriptor for the image in the memory.

Abstract: A method for receiving a first text in a source language is provided. The method includes associating the first text in the source language with a first vector, selected from a plurality of vectors associated with the source language in an embedded set, and identifying a second vector that is closer to the first vector than a pre-selected threshold. The second vector is associated with a second text in a target language. The method includes associating the first text in the source language with the second text in the target language, wherein the embedded set includes a first image vector for a first image and a second image vector for a second image, and returning the second text in the target language, the second text being a translation of the first text from the source language to the target language. A system configured to perform the above method is also provided.

Abstract: A computer-implemented method is provided for retrieving an image from a user in a desired format and for detecting a compression efficiency for the image. When the compression efficiency is above a pre-selected threshold the computer-implemented method includes obtaining a saliency representation of the image, capturing a feature description of a non-salient portion of the image, flattening the non-salient portion in a new image, storing the new image in a selected format in a memory and storing a background descriptor for the image in the memory.