Abstract: An anti-PD-L1 antibody, or an antigen-binding fragment thereof, comprising: a heavy chain variable region comprising the three CDRs with the sequences of SEQ ID NOs: 2-4, 6-8, 10-12, 14-16, or 18-20; and/or a light chain variable region comprising the three CDRs with the sequences of SEQ ID NOs: 22-24, 26-28, 30-32, 34-36, or 38-40, wherein the antibody is a chimeric, humanized, composite, or human antibody.

Abstract: The present disclosure provides a virus composition. The efficacy of a variety of virus compositions loaded with different functional genes was detected, and several virus compositions with excellent antitumor effect were confirmed.

Abstract: A data processing system includes: a processor; a memory storing executable instructions which, when executed by the processor, cause the processor, alone or in combination with other processors, to implement a client application with a user interface. The client application is configured to: receive an image depicting an object; generate a fine-tuning input to an image-generating Artificial Intelligence (AI) model to associate image data of the object with an identifier; with the fine-tuning input, fine-tune the AI model; structure a prompt for the AI model using the identifier; and obtain from the AI model a new customized image that depicts the object while preserving an appearance of the object.

Abstract: A system for cropping an image is disclosed, which performs receiving a source image and user intention data; determining a target feature based on the user intention data; identifying a plurality of visual features within the source image; determining a contextual relevance between the target feature and each identified visual feature of the source image; identifying, based on the determined contextual relevance between the target feature and each identified visual feature of the source image, one or more cropping candidate portions within the source image; cropping, based on the one or more cropping candidate portions, the source image to generate a plurality of cropped images; and causing the plurality of cropped images to be displayed on a display.

Abstract: A data processing system for requesting a customized image from an image-generating artificial intelligence engine may include a processor and a memory comprising instructions for execution by the processor. The instructions, when executed by the processor, cause the processor to: accept user input from a user, the user input comprising an image; tokenize the image to generate a set of tokens for use by the image-generating artificial intelligence engine; and submit the set of tokens to the image-generating artificial intelligence engine to support a request by the user for a customized image corresponding to the tokenized image.

Abstract: A bidirectional portable energy storage power supply without an adapter includes an energy storage unit, a first full bridge circuit, a resonant network, a second full bridge circuit, a third full bridge circuit and a charging and discharging interface circuit connected in turn. Each of the first full bridge circuit, the second full bridge circuit and the third full bridge circuit can be used as an inverter circuit or a rectification circuit, the charging and discharging interface circuit switchably connected with a mains network and a workload, the resonant network combined with the first full bridge circuit or the second full bridge circuit to implement soft-switching. The present disclosure can implement to bidirectionally charge and discharge the portable energy storage power supply by omitting an external adapter thereof, to improve a charging and discharging conversion efficiency, shorten a charging time and reduce a volume of power supply.

Abstract: A system and method and for retrieving one or more visual assets includes receiving a search query for the one or more visual assets, the search query including textual data, encoding the textual data into one or more text embedding representations via a trained text representation machine-learning (ML) model, transmitting the one or more text embedding representations to a matching and selection unit, providing visual embedding representations of one or more visual assets to the matching and selection unit, comparing, by the matching and selection unit, the one or more text embedding representations to the visual embedding representations to identify one or more visual asset search results, and providing the one or more visual asset search results for display.

Abstract: A computer-implemented method, system, and computer program product for detecting or predicting the occurrence of critical heat flux in cooling systems. A model is built and trained to detect or predict the occurrence of critical heat flux in a cooling system based on acoustic data and/or image data. Such a model may be utilized to detect or predict the occurrence of critical heat flux in a cooling system by converting acoustic signals received from sensors remotely located from the cooling system from the temporal domain to a frequency domain. Furthermore, such a model may be utilized to detect or predict the occurrence of critical heat flux in a cooling system by extracting features from image data and acoustic data obtained from sensors remotely located from the cooling system and then concatenating such features.

Abstract: A computing system including a plurality of processing devices configured to execute a Mixture-of-Experts (MoE) layer included in an MoE model. The processing devices are configured to, in each of a plurality of iterations, at each of the processing devices, receive a respective plurality of input tokens. Executing the MoE layer further includes, at each of the processing devices, selecting one or more destination expert sub-models associated with the input tokens. Respective numbers k of expert sub-models selected differ across the iterations. At each of the processing devices, executing the MoE layer further includes conveying the input tokens to the one or more destination expert sub-models. Executing the MoE layer further includes generating one or more respective expert sub-model outputs at the one or more destination expert sub-models. Executing the MoE layer further includes generating and outputting an MoE layer output based on the one or more expert sub-model outputs.

Abstract: A computing system is provided, including a plurality of processing devices configured to execute a Mixture-of-Experts (MoE) layer included in an MoE model. The MoE layer includes a plurality of expert sub-models that each have a respective plurality of parameter values. The MoE layer is configured to be switchable between a data parallel mode and an expert-data-model parallel mode without conveying the respective parameter values of the expert sub-models among the plurality of processing devices.

Abstract: A computing system including a plurality of processing devices configured to execute a Mixture-of-Experts (MoE) layer included in an MoE model. The processing devices are configured to execute the MoE layer at least in part by, during a first collective communication phase between the processing devices, splitting each of a plurality of first input tensors along a first dimension to obtain first output tensors. Executing the MoE layer further includes processing the first output tensors at a respective a plurality of expert sub-models to obtain a plurality of second input tensors. Executing the MoE layer further includes, during a second collective communication phase between the processing devices, receiving the second input tensors from the expert sub-models and concatenating the second input tensors along the first dimension to obtain second output tensors. Executing the MoE layer further includes outputting the second output tensors as output of the MoE layer.

Abstract: According to implementations of the subject matter described herein, a solution is proposed for three-dimensional (3D) object detection. In this solution, feature representations of a plurality of points are extracted from point cloud data related to a 3D object. Initial feature representations of a set of candidate 3D objects are determined based on the feature representations of the plurality of points. Based on the feature representations of the plurality of points and the initial feature representations of the set of candidate 3D objects, a detection result for the 3D object is generated by determining self-correlations between the set of candidate 3D objects and cross-correlations between the plurality of points and the set of candidate 3D objects. In this way, without grouping points into candidate 3D objects, the 3D object in a 3D scene can be localized and recognized based on the self-correlations and cross-correlations.

Abstract: A scheme for modifying an image is disclosed, which includes receiving a source image having a first image configuration; determining a second image configuration for a target image; providing the received source image to an AI engine trained to identify, based on a set of rules related to visual features, candidate regions from the source image; generating proposal images based on the candidate regions, respectively; determining, based on prior aesthetical evaluation data, an aesthetical value of each regional proposal image; and selecting, based on the determined aesthetical value of each regional proposal image, one of the regional proposal images as the target image; extracting, from the AI engine, the target image; and causing the target image to be displayed via a display of a user device.

Abstract: An on-line detection system for defects of an MEA is provided. The detection system includes a workbench, two connecting rods are arranged inside the workbench, two ends of the two connecting rods are both connected to two side walls of the workbench by means of bearings, and conveying rollers are fixedly arranged outside the two connecting rods in a sleeving manner. One side of the workbench is fixedly provided with a first electric motor, and an output end of the first electric motor is fixedly connected to one of the connecting rods. Belt pulleys are fixedly arranged outside the two connecting rods in a sleeving manner, and a belt is arranged outside the two belt pulleys in a sleeving manner. A hollow roller is arranged on an inner side of the workbench, and a plurality of exhaust holes are provided in a top of the hollow roller.

Abstract: A sealing cover with quick-plug interface, comprising: cover body, above which a through hole is arranged, and in which a communication hole is arranged, and said through hole is connected with said communication hole; an annular sealing member and an annular fixing sleeve are arranged in said through hole, and said sealing member is arranged at the bottom of said through hole, and said fixing sleeve is arranged above said sealing member; said sealing member and said fixing sleeve are both clipped to an inner wall of the through hole, and said sealing member is made of a flexible material. There is no metal material in contact with the fluid inside the sealing cover with quick-plug interface, so there is no hidden danger of corrosion in long-term use, and the defect that the metal material is not resistant to corrosive fluids is avoided.

Abstract: A computing system including a plurality of processing devices configured to execute a Mixture-of-Experts (MoE) layer. The processing devices are configured to execute the MoE layer at least in part by receiving an input tensor including input tokens. Executing the MoE layer further includes computing a gating function output vector based on the input tensor and computing a sparse encoding of the input tensor and the gating function output vector. The sparse encoding indicates one or more destination expert sub-models. Executing the MoE layer further includes dispatching the input tensor for processing at the one or more destination expert sub-models, and further includes computing an expert output tensor. Executing the MoE layer further includes computing an MoE layer output at least in part by computing a sparse decoding of the expert output tensor. Executing the MoE layer further includes conveying the MoE layer output to an additional computing process.

Abstract: The present disclosure relates to methods and systems for generating a channel for use as a spatio-temporal constraint for motion planning for an autonomous vehicle. The method generates a triangulation mesh for a space in an environment. The triangulation mesh includes a plurality of nodes, each represents a geographic location in the space. Some of the nodes correspond to dynamic objects. Based on the triangulation mesh, a candidate channel extending from a starting location to a target location of the autonomous vehicle is generated. Further, the method predicts a time and location of a future triangulation mesh topology event caused by at least one of the nodes that will impact the candidate channel. A valid channel segment for the channel is selected, which extends from the starting location to a channel segment end location preceding the predicted location of the future triangulation mesh topology event. The channel improves the validity of motion planning for a longer time period.

Abstract: A consumption-free high-efficiency filter includes a tank body provided with a liquid outlet, a filter closure head provided with a liquid inlet and fixedly mounted with the tank body, a first group of support rings and second group of support rings that are respectively fixedly mounted at an upper end of the tank body and a lower end of the tank body, an upper cross support frame and a lower cross support frame, a mandrel, a first sliding bearing and a second sliding bearing used for rotatably mounting an upper end of the mandrel and a lower end of the mandrel, and a set of filter elements. Each filter element includes a filter element sleeve fixedly mounted on the mandrel, a blade and a filter loop.

Abstract: A data processing system implements a dynamic knowledge distillation process including dividing training data into a plurality of batches of samples and distilling a student model from a teacher model using an iterative knowledge distillation. The process includes instantiating an instance of the teacher model and the student model in a memory of the data processing system and obtaining a respective batch of training data from the plurality of batches of samples in the memory. The process includes training the teacher and student models using each of the samples in the respective batch of the training data, evaluating the performance of the student model compared with the performance of the teacher model, and providing feedback to student model to adjust the behavior of the student model based on the performance of the student model.

Abstract: Match-paired monoclonal antibodies against major royal jelly protein 4 (MRJP4) are secreted by hybridoma cell lines having microbial deposit numbers of CGMCC No. 17294 and CGMCC No. 17295, which are used in an ELISA kit and a colloidal gold immunoassay strip for detecting the MRJP4. The positive and MRJP4-specific cell lines are obtained by cell fusion using an antigen of MRJP4 recombinant protein and a cross-reaction with other major royal jelly proteins. The MRJP4 recombinant protein is used as an antigen to obtain several positive cell lines by cell fusion and two MRJP4-specific fusion cell lines are obtained by a cross-reaction with other major royal jelly proteins. Primary screening of a matched antibody pair is performed according to antibody pairing for recognizing different epitopes.