Abstract: An approach to real-time multi-agent visual question answering in a decentralized environment. The approach can include initializing a plurality of agents based on data associated with an image and a question about the image. The approach can include identifying a first portion of the plurality of agents to construct a grid of agents based on similarity between the first portion of agents. The approach can include determining a second portion of agents of the first portion of agents with edges between agents allowing sharing of agent concepts and agent parameters. The approach can include generate training trajectories of the grid of agents based on executing a collaborative rollout. Additionally, the approach can include optimizing agent policies of the grid of agents.

Abstract: The embodiments are directed to generating source code for a program from a problem description. One or more pre-trained code large language models (LLMs) generate sub-modules from a problem description in a natural language. The sub-modules are filtered based on testing criteria and encoded into sub-module encodings in an embedding space. The sub-module encodings are clustered into multiple clusters. A subset of sub-modules encoding that are close to the centroids of the clusters are selected. The sub-set of sub-modules is decoded into representative sub-modules. The problem description is augmented with the representative sub-modules and fed into one or more pre-trained code LLMs and new sub-modules are generated. The iterations continue until a program is generated from the representative sub-modules.

Abstract: Embodiments described herein provide training a prompt generator for text classification. A first training dataset associated with a first plurality of class labels is received for a first training process. For a first instance of the first training dataset, a set of labels of interest is generated by sampling from a set of possible class labels including the first plurality of class labels. The prompt generator generates a first prompt based on the set of labels of interest. A pretrained language model generates a task output in response to an input of the first instance prepended with the first prompt. A loss objective is generated based on the task output and the set of labels of interest. Parameters of the prompt generator are updated based on the computed loss function via backpropagation while the pretrained language model is frozen.

Abstract: Embodiments described herein provide a system for training a neural network model using a teacher-student framework. The system includes a communication interface configured to communicate with a teacher model; a memory storing a student model and a plurality of processor-executable instructions; and a processor executing the processor-executable instructions to perform operations. The operations include: generating, by the student model, a first task output in response to a task input; obtaining, from an evaluation environment, a feedback relating to an accuracy of the first task output; obtaining a refinement output generated by the teacher model based on an input of the first task output and the feedback; and training the student model based on a training input of the first task output and the feedback and a training label of the refinement output.

Abstract: Embodiments described herein provide a soft prompt tuning technique referred to as the Vector quantized Input-contextualized Prompt (VIP). The VIP techniques has two integral properties i) instead of learning a fixed set of prompt tokens irrespective of the input, it generates a contextualized version of the soft prompts, conditional on the input text ii) it further passes the input-contextualized prompt tokens through a quantization network, inspired by Vector Quantized Transformers. The quantization network uses nearest neighbor search over a learnable codebook to train a discrete latent variable model over the prompt-space, thus generating quantized version of contextual prompt tokens. These quantized contextual prompt tokens are finally fed into the frozen language model along with the original input text.

Abstract: Methods, apparatus, and processor-readable storage media for facilitating a secure platform for point-to-point brain sensing are provided herein. An exemplary method includes presenting a user of a brain-computer interface with learning content; monitoring brain signals of the user using a brain-computer interface while the learning content is being presented; determining, based at least in part on said monitoring, that a confusion state of the user exceeds a personalized confusion threshold in regard to at least a part of the learning content; in response to said determining, outputting information to assist the user in understanding the part of the learning content until the confusion state of the user is below the personalized confusion threshold.

Abstract: Embodiments described herein provide a data driven framework that (i) translates demonstration examples to a fixed-length soft prompt—a sequence of soft tokens; and (ii) learns a global (not generated from demonstrations) soft prompt. The framework then combines the global prompt, the translated prompts and the original context to create an augmented context which is given as final input for the backbone LM to use.

Abstract: Methods, systems, apparatuses, devices, and computer program products are described. A system may collect a first set of profiling data associated with computational resource consumption of one or more code implementations or methods. The system may use a vector embedding translation to convert the profiling data into one or more vector spaces. Each vector space may include a set of vectors, and each vector may correspond to an execution of a code implementation or method. The system may use the vector spaces to generate a model representation of the computational resource consumption of the one or more code implementations. In some cases, the system may collect and convert a second set of real-time profiling data into vector spaces, which the system may compare to the model representation such that users may identify deviations from resource consumption footprints.

Abstract: Embodiments described herein provide training a prompt generator for text classification. A first training dataset associated with a first plurality of class labels is received for a first training process. For a first instance of the first training dataset, a set of labels of interest is generated by sampling from a set of possible class labels including the first plurality of class labels. The prompt generator generates a first prompt based on the set of labels of interest. A pretrained language model generates a task output in response to an input of the first instance prepended with the first prompt. A loss objective is generated based on the task output and the set of labels of interest. Parameters of the prompt generator are updated based on the computed loss function via backpropagation while the pretrained language model is frozen.

Abstract: Embodiments described herein provide label modular prompts for a text classification task. A label modular prompt generator may determine a set of class labels of interest from a set of possible class labels associated with an input text sequence. The label modular prompt generator may generate a plurality of label prompts based on the set of class labels of interest. A first class label and a sequence of soft tokens that are generated based on representations associated with the first class label are concatenated into a first label prompt. The soft tokens are tunable using a plurality of parameters of the label modular prompt generator. The label modular prompt generator may provide an input of the input text sequence prepended with the plurality of label prompts to a pretrained language model. The pretrained language model may generate a task output in response to the input text sequence.

Abstract: Some embodiments of the current disclosure disclose methods and systems for analyzing root causes of an incident disrupting information technology services such as cloud services. In some embodiments, a set of problem review board (PRB) documents including information about said incidents may be parsed using a natural language processing (NLP) neural model to extract structured PRB data from the unstructured investigative information contained in the PRB documents. The structured PRB data may include symptoms of the incident, root causes of the incident, resolutions of the incidents, etc., and a causal knowledge graph causally relating the symptoms, root causes, resolutions of the incidents may be generated.

Abstract: Embodiments described herein provide a soft prompt tuning technique referred to as the Vector quantized Input-contextualized Prompt (VIP). The VIP techniques has two integral properties i) instead of learning a fixed set of prompt tokens irrespective of the input, it generates a contextualized version of the soft prompts, conditional on the input text ii) it further passes the input-contextualized prompt tokens through a quantization network, inspired by Vector Quantized Transformers. The quantization network uses nearest neighbor search over a learnable codebook to train a discrete latent variable model over the prompt-space, thus generating quantized version of contextual prompt tokens. These quantized contextual prompt tokens are finally fed into the frozen language model along with the original input text.

Abstract: Embodiments described herein provide a soft prompt tuning technique referred to as the Vector quantized Input-contextualized Prompt (VIP). The VIP techniques has two integral properties i) instead of learning a fixed set of prompt tokens irrespective of the input, it generates a contextualized version of the soft prompts, conditional on the input text ii) it further passes the input-contextualized prompt tokens through a quantization network, inspired by Vector Quantized Transformers. The quantization network uses nearest neighbor search over a learnable codebook to train a discrete latent variable model over the prompt-space, thus generating quantized version of contextual prompt tokens. These quantized contextual prompt tokens are finally fed into the frozen language model along with the original input text.

Abstract: Some embodiments of the current disclosure disclose methods and systems for analyzing root causes of an incident disrupting information technology services such as cloud services. In some embodiments, a set of problem review board (PRB) documents including information about said incidents may be parsed using a natural language processing (NLP) neural model to extract structured PRB data from the unstructured investigative information contained in the PRB documents. The structured PRB data may include symptoms of the incident, root causes of the incident, resolutions of the incidents, etc., and a causal knowledge graph causally relating the symptoms, root causes, resolutions of the incidents may be generated.

Abstract: Methods, systems, and computer program products for pressure-based apparel image searching are provided herein. A computer-implemented method includes converting images in a product catalog of an electronic commerce website to a predetermined representation; storing the converted images in an index; determining a first object of interest within an image derived from a social media post and displayed on a screen, by detecting physical contact imparted by a user at a position on the screen corresponding to where the first object of interest is located; quantifying the amount of pressure applied by the user via the physical contact; determining additional objects of interest within the image based on the amount of pressure applied by the user; retrieving, from the index, images of products corresponding to the first object of interest and images of products corresponding to the additional objects of interest; and displaying the retrieved images on the screen.

Abstract: An approach to real-time multi-agent visual question answering in a decentralized environment. The approach can include initializing a plurality of agents based on data associated with an image and a question about the image. The approach can include identifying a first portion of the plurality of agents to construct a grid of agents based on similarity between the first portion of agents. The approach can include determining a second portion of agents of the first portion of agents with edges between agents allowing sharing of agent concepts and agent parameters. The approach can include generate training trajectories of the grid of agents based on executing a collaborative rollout. Additionally, the approach can include optimizing agent policies of the grid of agents.

Abstract: One embodiment provides a method, including: receiving, at an information handling device, drawing input; identifying, using a processor, at least one object in the drawing input; determining, based on the identifying, whether a factual anomaly exists in the drawing input with respect to the at least one object; and notifying, responsive to determining that a factual anomaly exists, a user of the factual anomaly.

Abstract: Embodiments described herein provide systems and methods for a partially supervised training model for questioning answering tasks. Specifically, the partially supervised training model may include two modules—a query parsing module and a program execution module. The query parsing module parses queries into a grogram, and the program execution module execute the program to reach an answer through explicit reasoning and partial supervision. In this way, the partially supervised training model can be trained with answers as supervision, obviating the need for supervision by gold program operations and gold query-span attention at each step of the program.

Abstract: Embodiments describing an approach to evaluate text and image consistency. Receiving one or more images. Receiving one or more text documents. Identifying relevant text in the one or more text documents. Determining the consistency between the one or more images and the one or more text documents. Creating one or more image and text consistency scores based on the determined consistency between the one or more images and the one or more text documents, and outputting the one or more image and text consistency scores for evaluating text and image consistency.

Abstract: One embodiment provides a method, including: detecting, on a social media application, a media object comprising at least one image of a final product made from at least one purchased raw product; extracting, from the media object and text corresponding to the social media application post, (i) information related to the final product and (ii) social feedback regarding the final product; determining, based upon the social feedback, that the use of the at least one purchased raw product into the final product comprises positive social feedback; and providing a recommendation regarding the use of the at least one purchased raw product to one or more other individuals, wherein the recommendation is generated using the information related to the final product.