Abstract: This application provides an inference method and a communication apparatus. In this method, a first AI model may obtain, through inference based on a first prompt entered by a user and local information of a network device in a geographical range related to the first prompt, a second prompt including context information of the first prompt, so that the first AI model or a second AI model can perform inference based on the first prompt with the context information supplemented, to obtain a more accurate reply result.

Abstract: A distributed training method for an artificial intelligence AI model and an apparatus are provided. A subnode that participates in training of a model of a central node feeds back a representation difference between models to the central node, and the central node issues model training indication based on the representation difference, so that a high-performance machine learning model can be obtained within a limited quantity of communication processes between the subnode and the central node, to improve performance of distributed training.

Abstract: A diode structure includes a nanosheet structure on a substrate, including a first, first-type diffusion region, a second, first-type diffusion region on the substrate, a first, second-type diffusion region, and a second, second-type diffusion region, each on the substrate. The diode structure includes a first gate on the nanosheet structure between the first and second, first-type diffusion regions. The diode structure includes a first frontside zero (M0) metal layer coupled to a frontside of the first and second, first-type diffusion regions, and a first backside M0 metal layer coupled to a backside of the first and second, first-type diffusion regions to form an anode. The diode structure includes a second frontside M0 metal layer coupled to a frontside of the first and second, second-type diffusion regions, and a second backside M0 metal layer coupled to a backside of the first and second, second-type diffusion regions to form a cathode.

Abstract: Disclosed are complementary field effect transistor (CFET) circuits with vertical routing structures and methods for making the same. In an aspect, a semiconductor structure comprises a first field effect transistor (FET) of a first charge carrier type, comprising a first source/drain (S/D) region, a second S/D region, and a first gate; a second FET of a second charge carrier type, disposed above the first FET and comprising a third S/D region, a fourth S/D region, and a second gate; a frontside metal (FM) layer disposed above the second FET and comprising an FM conductor extending in an X direction; a backside metal (BM) layer disposed below the first FET and comprising a BM conductor extending in the X direction; and a vertical connector extending in the Z direction, that electrically couples the BM conductor to the FM conductor.

Abstract: The provided is a system for fully automatically implementing automated liquid dispensing on multiple relay tanks. The system includes a plurality of concentration preparation tanks, a plurality of concentration tanks, a water-for-injection filling tank, an online dilution system and a plurality of relay tanks. Each concentration preparation tank is connected to a plurality of the concentration tanks, the number of the concentration tanks connected to each concentration preparation tank being same; each concentration tank and the water-for-injection filling tank are both connected to each relay tank via the online dilution system. The system for fully automatically implementing automated liquid dispensing on multiple relay tanks can further reduce the sizes of the relay tanks, thereby reducing the whole area for buffer solution preparation, and satisfying the demand of short time for customers.

Abstract: This disclosure provides a communication method, which includes: obtaining M pieces of channel estimation information, where an mth piece of channel estimation information in the M pieces of channel estimation information is estimation information of a channel between an mth cell node in M cell nodes and a terminal device; determining a type of a target positioning manner based on the M pieces of channel estimation information; and obtaining N pieces of channel estimation information based on the type of the target positioning manner, where the N pieces of channel estimation information are used to position the terminal device, an nth piece of channel estimation information in the N pieces of channel estimation information is estimation information of a channel between an nth cell node in N cell nodes and the terminal device, the N cell nodes are included in the M cell nodes.

Abstract: Embodiments concern methods and composition for preventing or treating a bacterial infection, particularly infection by a Staphylococcus bacterium. The embodiments concern improved non-toxigenic Protein A (SpA) variant.

Abstract: An ultrasonic non-destructive testing method and system for detecting lithium precipitation in lithium-ion batteries are provided. The method includes the following steps: acquiring full matrix capture data of the lithium-ion battery using an ultrasonic phased array transducer; inputting the measured ultrasonic data into the initial fully convolutional network (FCN), which outputs a predicted ultrasonic velocity model; using the predicted model as the initial input for an acoustic wave equation that outputs reconstructed ultrasonic data; constructing a loss function using the deviation between the measured and reconstructed ultrasonic data, and iterating until the loss function converges. This process completes the training of the unsupervised FCN. The trained FCN then outputs an accurate reconstructed ultrasonic velocity model that represents the lithium precipitation in the battery.

Abstract: This disclosure relates to the field of localization technologies, and provides a communication method and a communication apparatus. In the method, a core network device obtains at least one piece of location information, where the at least one piece of location information is from different apparatuses, and each piece of location information identifies a location of a same reflecting surface on a spatial map. The core network device determines, from the at least one piece of location information, location information with highest confidence or location information with confidence higher than a value, and constructs, based on the location information, a fingerprint database for localization. In this way, accuracy of locating a terminal device can be effectively improved.

Abstract: This application provides a communication method and a communication apparatus. The method includes: A first communication apparatus compresses N pieces of second channel information based on X pieces of first channel information to obtain M pieces of compressed information. For example, an ith piece of compressed information is obtained by compressing Pi pieces of second channel information based on a jth piece of first channel information, and obtaining time of the first channel information is earlier than obtaining time of the second channel information. The first communication apparatus sends the ith piece of compressed information, and the first communication apparatus further sends the jth piece of first channel information. A second communication apparatus restores the Pi pieces of second channel information based on the ith piece of compressed information and the jth piece of first channel information that may be received at different time. Feedback overheads can be further reduced.

Abstract: A chip includes a first diffusion region extending in a first direction, and a second diffusion region extending in the first direction, wherein the first diffusion region and the second diffusion region are stacked in a second direction perpendicular to the first direction. The chip also includes a first rail extending in the first direction, wherein the first diffusion region and the first rail are spaced apart in a third direction perpendicular to the first direction and the second direction. The chip also includes a third diffusion region extending in the first direction, wherein the third diffusion region is wider than the first diffusion region in the third direction, and a fourth diffusion region extending in the first direction, wherein the third diffusion region and the fourth diffusion region are stacked in the second direction.

Abstract: Example uplink precoding methods and apparatuses are provided. One example method includes receiving a first reference signal from a terminal device. A first codebook is determined based on the received first reference signal, wherein the first codebook is used to precode an uplink signal. The first codebook is then sent to the terminal device. In a second example, a first reference signal is sent to an access network device. A first codebook is then received from the access network device and an uplink signal is precoded using the first codebook.

Abstract: A chip includes a first source/drain, a second source/drain, a top gate disposed between the first source/drain and the second source/drain, and one or more first channels coupled between the first source/drain and the second source/drain, wherein the one or more first channels pass through the top gate. The chip also includes a third source/drain, wherein the third source/drain is vertically stacked with the first source/drain, a fourth source/drain, wherein the fourth source/drain is vertically stacked with the second source/drain, and a first insulator disposed below the top gate and between the third source/drain and the fourth source/drain, wherein the first insulator electrically isolates the top gate from the third source/drain and the fourth source/drain.

Abstract: This application relates to the field of mobile communication, and in particular, to a communication method and apparatus, and a readable storage medium, and may be applied to an NR protocol framework, a 6G protocol framework, or the like. The method includes: A UE obtains N pieces of channel information, and sends first information to a base station, where the first information indicates a compression and feedback method for the N pieces of channel information, and the compression and feedback method includes one or both of the following: a quantity of pieces of channel feedback information or a correspondence between channel feedback information and channel information, where one piece of channel information corresponds to one or more pieces of channel feedback information. According to embodiments of this application, flexibility of channel information feedback can be improved.

Abstract: The present application provides an antibiotic resistance gene-free miniplasmid, comprising a nucleotide sequence encoding an antitoxin protein, and a replicon. A backbone of the antibiotic resistance gene-free miniplasmid of the present application is smaller and controlled within 1,000 bp, redundant useless fragments are reduced, and meanwhile, the miniplasmid does not comprise a resistance expression cassette, thereby improving the utilization rate of a sequence of interest in the plasmid and reducing the production burden. The miniplasmid has higher efficiency and safety in the plasmid DNA-mediated non-viral vector delivery, gene therapy, vaccine immunity, virus production, antibody production, and other genetic engineering.

Abstract: This application provides an artificial intelligence model training method and a communication apparatus, and relates to the field of communication technologies. In the artificial intelligence model training method, a parameter that needs to be updated in a first model may be determined based on obtained recognition information of a first scenario, and the parameter that needs to be updated in the first model is updated, but not all parameters are updated, to effectively reduce overall overheads for model update.

Abstract: A chimeric lysozyme variant has an improved heat resistance and is encoded by a polynucleotide. A novel chimeric lysozyme sequence is constructed, and high-efficiency expression in a host cell is achieved. The thermal stability of the chimeric lysozyme under the condition of 60° C. is nearly doubled compared to a parental sequence. The chimeric lysozyme variant can be used to prepare product that is antibiotic, for example, animal feed additive containing the chimeric lysozyme variant and an animal feed containing the animal feed additive.

Abstract: An image processing method based on a dither algorithm and a display device are disclosed. The method includes: obtaining a to-be-displayed image, and determining a first region and a second region of the to-be-displayed image, where the first region and the second region are determined according to a coverage range of a visual field of a user on a display screen; determining a first dither algorithm corresponding to the first region and a second dither algorithm corresponding to the second region, where parameter matrices the first dither algorithm and the second dither algorithm are different; and obtaining a rendered image of the to-be-displayed image by rendering the first region according to the first dither algorithm and rendering the second region according to the second dither algorithm, and displaying the rendered image on the display screen.

Abstract: Provided is a chimeric antigen receptor. An antigen binding domain thereof comprises a mesothelin binding molecule containing an anti-mesothelin single-domain antibody. Immune cells containing the chimeric antigen receptor described herein have a higher cell killing effect.