Abstract: Techniques for multilingual batch alignment training and/or fine-tuning for language models are included herein. The system generates and/or accesses data points from batches of data having semantically equivalent content across languages and/or semantically equivalent prompts across languages. The prompts include queries related to a topic or a predictive language task. The data points include one or more sets of resources that are semantic equivalents in different languages. Data points for which a resource enables a correct response are used as positive training data points for reinforcement learning. Data points for which a resource does not enable a correct response are considered as negative training data points for reinforcement learning. The data points are used for alignment training and/or other reinforcement learning for generative language models.

Abstract: A method may include receiving a primary document including one or more selection boxes, one or more text lines, and one or more annotations. The method may include determining, a class based on the annotations. The method may include identifying the one or more selection boxes and one or more text lines of the primary document. The method may include generating a graph representing the one or more selection boxes and the one or more text lines. The method may include mapping each of the one or more selection boxes to a respective text line of the one or more text lines of the graph based at least in part on one or more characteristics associated with the selection boxes. The method may include generating a key-value pair associated with each of the one or more text lines and generating a document model of the primary document.

Abstract: Techniques for evaluating a language model are disclosed herein. One or more agents and/or one or more language models are used to generate responses to queries in conversations. For example, during chat session conversations, the agent and/or model used to generate responses to queries in the conversation is sometimes changed. Various metrics are collected for the conversations for which a prompt, agent or model change occurs and/or the conversations for which a change to the prompt, agent and/or model does not occur. Metrics such as success rate, escalation rate, query repetition, direct feedback, or other feedback is collected and/or aggregated for groups of users and/or topics of conversations. One or more models used to generate responses in the conversations are evaluated and/or selected based on the metrics.

Abstract: In some implementations, techniques described herein may include identifying text in a visually rich document and determining a sequence for the identified text. The techniques may include selecting a language model based at least in part on the identified text and the determined sequence. Moreover, the techniques may include assigning each word of the identified text to a respective token to generate textual features corresponding to the identified text. The techniques may include extracting visual features corresponding to the identified text. The techniques may include determining positional features for each word of the identified text. The techniques may include generating a graph representing the visually rich document, each node in the graph representing each of the visual features, textual features, and positional features of a respective word of the identified text. The techniques may include training a classifier on the graph to classify each respective word of the identified text.

Abstract: Techniques for evaluating a language model are disclosed herein. A language model is used to generate responses to queries using documents as context. A model evaluator evaluates a query, document, response, triplet to classify the triplet. The triplets are evaluated using key-phrase extraction, summary generation, cross-encoding and similarity scoring for members of the triplets. The triplets are evaluated by generating a summary of the document and evaluating similarity for the summary and response. The triplets are aggregated and/or clustered to determine one or more classifications or attributes based on the scores. Classifications and/or attributes of triplets are used as training data to train and/or fine tune the language model. A classification of a triplet is used to indicate that a chatbot should switch from a first language model to a second language model for generating a subsequent response to a subsequent query.

Abstract: A method for a fully distributed privacy-preserving multi-lateral netting between parties is disclosed. The method comprising collecting pending transactions between at least a part of the parties, publishing a commitment of pending transactions to a bulletin board, secretly sharing de-commitments among the parties, and publishing, by each party, its determined liquidity status, to the bulletin board. Furthermore, the method comprises upon determining a failure in the publishing of any liquidity status, performing the following MPC by determining whether all pending transactions are settable maintaining positive liquidity for each party, and upon determining a failure in the determining, for each party, whether all pending transactions are nettable determining an optimized subset of transactions that are nettable without violating liquidity requirements for each party.

Abstract: Techniques for responding to multilinguistic queries using an agent core are disclosed herein. An agent core is trained and/or fine-tuned in a first language to generate instructions (i.e., commands) for answering a query. A language identification and/or translation model receives queries, identifies languages associated with the queries, and translates the queries to the first language. The agent core generates instructions in the first language based on the translated queries. The instructions include instructions, in the first language, to perform actions such as retrieval, generation, contextual understanding, or calculation, in both the first language and the second language. The results of executing the instructions, including an action performed in the first language and an action performed in one or more second languages, are combined, ranked and/or reranked to generate an answer to the query.

Abstract: Some embodiments include an apparatus having a flip-flop circuit, which can include a first tristate inverter, a second tristate inverter including an input node coupled to an input node of the first tristate inverter; a first additional inverter including, and a second additional inverter including an output node coupled to an output node of the first additional inverter; a first memory including a first memory node coupled to an output node of the second tristate inverter, and a first additional memory node coupled to an input node of the first additional inverter; and a second memory including a second memory node coupled to an output node of the first tristate inverter, and a second additional memory node coupled to an input node of the second additional inverter.

Abstract: A method may include detecting one or more selection boxes and one or more text lines in a primary document. The method may include determining respective vectors associated with the selection box and adjacent text lines to the selection box in a plurality of directions. The method may include determining a set of respective vectors associated with a unique selection box. The method may include determining a variance between respective vectors in the set of respective vectors and identifying a particular direction corresponding to a minimal variance between the respective vectors in the set of respective vectors as compared to a variance of other sets of respective vectors. The method may include generating a key-value pair based on the set of respective vectors characterized by the minimal variance. The method may include generating a document model, including the key-value pair, and extracting data according to the document model.

Abstract: Techniques and mechanisms for an integrated clock gate (ICG) to selectively output a clock signal, and to provide frequency division functionality. In an embodiment, an ICG circuit comprises first circuitry which is coupled to receive a first clock signal, and second circuitry which is coupled to receive a control signal. The first circuitry provides a single edge triggered flip-flop functionality, and is coupled to communicate a feedback signal which the first circuitry is further coupled to receive. Based on the control signal and the feedback signal, the second circuitry performs an exclusive OR (XOR) operation to selectively enable the first circuitry to generate a second clock signal based on the first clock signal. In another embodiment, a frequency of the second clock signal is substantially equal to one half of a frequency of the first clock signal.

Abstract: Embodiments described herein are directed towards a synthetic document generation pipeline for training artificial intelligence models. One embodiment includes a method including a device that receives an instruction to generate a document to be used as a training instance for a first machine learning model, the instruction including an element configuration, a document class configuration, a format configuration, an augmentation configuration, and data bias and fairness. The device can receive an element from an interface based at least in part on the element configuration, the element can simulate a real-world image, real-world text, or real-world machine-readable visual code. The device can generate metadata describe a layout for the element on the document based on the document class configuration. The device can generate the document by arranging the element on the document based on the metadata, wherein the document is generated in a format based on the format configuration.

Abstract: A system and method for automatically transitioning between carrier and IP messaging that can include receiving a messaging request specifying a destination endpoint for a message, retrieving client status of the destination endpoint from a client application registry system, determining, based on whether the client application is reachable using an IP-based communication protocol and on features supported by the client application, a messaging protocol option selected from one or more carrier-based communication options and one or more IP-based communication options, and transmitting the message to the destination endpoint using the selected messaging protocol option.

Abstract: In some implementations, techniques described herein may include identifying text in a visually rich document and determining a sequence for the identified text. The techniques may include selecting a language model based at least in part on the identified text and the determined sequence. Moreover, the techniques may include assigning each word of the identified text to a respective token to generate textual features corresponding to the identified text. The techniques may include extracting visual features corresponding to the identified text. The techniques may include determining positional features for each word of the identified text. The techniques may include generating a graph representing the visually rich document, each node in the graph representing each of the visual features, textual features, and positional features of a respective word of the identified text. The techniques may include training a classifier on the graph to classify each respective word of the identified text.

Abstract: A data processing system implements obtaining a first textual content, segmenting the first textual content into a plurality of first segments, and providing each segment of the plurality of first segments to a first natural language processing (NLP) model to obtain a set of first readability scores for the plurality of first segments. The first NLP model is configured to analyze a textual input and to output a readability score representing a measurement of readability of the textual input. The system further implements aggregating the set of first segment readability scores to determine a first readability score for the first textual content, and perform at least one of causing the first readability score to be presented to a user or performing one or more actions on the first textual content based on the readability score.

Abstract: The present disclosure is directed to systems and methods of providing content. A server can generate a request for a push content item for an account identifier linked with a computing device. The server can establish a push auction for the account identifier with multiple candidate push content items. The server can determine an auction score for each candidate push content item and select a push content item therefrom based on the auction score. The server can determine a parameter for the account identifier and control delivery of the selected push content item based on a delivery control policy. The server can compare a value of the parameter with a threshold value to authorize the push content item. The server can provide the selected and authorized push content item for presentation in a push content slot via the computing device linked to the account identifier.

Abstract: Techniques for hard negative mining for ranking models are provided. In one technique, an input document that is associated with a query is received and input to an embedding model, which outputs a document embedding (DE). Based on the document embedding, multiple embeddings are identified. Clusters of embeddings are generated from the multiple embeddings. A cluster that includes the DE is identified. Based on the DE, two sets of embeddings are identified in the cluster. For each embedding in a first set of embeddings: (1) a particular embedding (PE) is selecting from the second set of embeddings based on a similarity score between the embedding and the PE; (2) a first document that is associated with the PE is identified; and (3) a training instance that includes the first document is generated and added to training data. A model is trained based on the training data.

Abstract: A memory array of a compute tile may store activations or weights of a DNN. The memory array may include databanks for storing contexts, context MUXs, and byte MUXs. A databank may store a context with flip-flop arrays, each of which includes a sequence of flip-flops. A logic gate and an ICG unit may gate flip-flops and control whether states of the flip-flops can be changed. The data gating can prevent a context not selected for the databank from inadvertently toggling and wasting power A context MUX may read a context from different flip-flop arrays in a databank based on gray-coded addresses. A byte MUX can combine bits from different bytes in a context read by the context MUX. The memory array may be implemented with bit packing to reduce distance between the context MUX and byte MUX to reduce lengths of wires connecting the context MUXs and byte MUXs.

Abstract: Embodiments described herein are directed towards a synthetic document generation pipeline for training artificial intelligence models. One embodiment includes a method including a device that receives an instruction to generate a document to be used as a training instance for a first machine learning model, the instruction including an element configuration, a document class configuration, a format configuration, an augmentation configuration, and data bias and fairness. The device can receive an element from an interface based at least in part on the element configuration, the element can simulate a real-world image, real-world text, or real-world machine-readable visual code. The device can generate metadata describe a layout for the element on the document based on the document class configuration. The device can generate the document by arranging the element on the document based on the metadata, wherein the document is generated in a format based on the format configuration.

Abstract: A method may include detecting one or more selection boxes and one or more text lines in a primary document. The method may include determining respective vectors associated with the selection box and adjacent text lines to the selection box in a plurality of directions. The method may include determining a set of respective vectors associated with a unique selection box. The method may include determining a variance between respective vectors in the set of respective vectors and identifying a particular direction corresponding to a minimal variance between the respective vectors in the set of respective vectors as compared to a variance of other sets of respective vectors. The method may include generating a key-value pair based on the set of respective vectors characterized by the minimal variance. The method may include generating a document model, including the key-value pair, and extracting data according to the document model.