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: Methods, apparatuses and systems for providing a switching component are disclosed herein. An example switching component may comprise: A switching component comprising: a housing; a carrier body disposed within the housing; a first pair of contact pads disposed on a first surface of the carrier body; and a second pair of contact pads disposed on a second surface of the carrier body, wherein each pair of contact pads is configured to independently make contact with adjacent bridge contact pads in order to actuate an electrical bridge in response to movement of the carrier body.

Abstract: System for free-space light coupling between 2-dimensional light transmitter and receiver arrays using flat optics. The system includes: a transmitter array configured to emit one or more light beams; a first flat optic configured to receive the light beams from the transmitter array and generate one or more intermediate light beams; a first spacer disposed between the light transmitter array and the first flat optic; a receiver array; a second flat optic configured to receive the intermediate light beams and generate one or more output light beams toward the receiver array; and a second spacer disposed between the second flat optic and the light receiver array. The TX and RX portions of the light coupling system may be used in LiDAR systems as the light emitter (beam steerer) and detector, respectively. The flat optics are configured to provide a wide field of view for the LiDAR.

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: 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: Various embodiments of the present disclosure are directed towards an integrated chip including a dielectric structure sandwiched between a first electrode and a bottom electrode. A passivation layer overlies the second electrode and the dielectric structure. The passivation layer comprises a horizontal surface vertically below a top surface of the passivation layer. The horizontal surface is disposed above a top surface of the dielectric structure.

Abstract: A tool includes a housing, a drive assembly including an output shaft extending from the housing such that a tool element for performing work on a workpiece is attachable to the output shaft, and a bushing disposed within the housing. The bushing rotatably supports the output shaft. The bushing is impregnated with lubricant such that the bushing is self-lubricating, and the bushing has an impregnation ratio between about 5% and about 25%.

Abstract: Meta-lens based ocular imaging, near-eye display, and eye-tracking systems are described. The systems can include a single focusing optic and an integrated circuit that provides illumination light and includes an imaging array. The focusing optic includes meta-atoms formed on a substrate. The systems may have no moving parts and achieve imaging or image-projection fields-of-view approaching or exceeding 180 degrees. Because of their low part count, the systems can be robust and have a very small form factor.

Abstract: A wide field-of-view sensor or projector includes a transparent substrate with one or more apertures on one side, and one or more chip stacks joined to the opposite side. Each chip stack includes a flat optics layer (e.g., metasurface), at least one spacer layer, an optional filter layer, and either an image sensor or a light source. In one example, two apertures and two corresponding chip stacks are provided; both chip stacks include image sensors but different metasurfaces and filters to capture different information from the scene. In an alternative embodiment, the two chip stacks include a light source and an image sensor, to function as a light projector and a light receiver, respectively. In other examples, two apertures and a single chip stack are provided, where the single chip stack include two metasurface and/or two filters corresponding to the two apertures.

Abstract: A mask for forming a trench in a flexible bendable region of a flexible display panel is provided. The mask includes a first region, a second region, and a third region sandwiched between the first and second regions in a first direction, and the third region has the same pattern as a pattern of a trench to be formed. Light transmission properties of the first and second regions are the same as each other, but are opposite to a light transmission property of the third region. An edge of at least one of the first and second regions proximal to the third region has a plurality of protrusions, and each of the plurality of protrusions has a vertex angle that is at a side proximal to the third region and is not more than 90°. A flexible display panel and a manufacturing method thereof are further provided.

Abstract: The disclosure provides a three-dimensional (3D) image classification method and apparatus, a device, and a storage medium. The method includes: obtaining a 3D image, the 3D image including first-dimensional image information, second-dimensional image information, and third-dimensional image information; extracting a first image feature corresponding to planar image information from the 3D image; extracting a second image feature corresponding to the third-dimensional image information from the 3D image; fusing the first image feature and the second image feature, to obtain a fused image feature corresponding to the 3D image; and determining a classification result corresponding to the 3D image according to the fused image feature corresponding to the 3D image.

Abstract: An electronic device (200) includes a liquid crystal display (LCD) panel (204), a cover layer (218), one or more light sources (224), and an optical sensor module (226). The LCD panel (204) receives a contact input associated with a fingerprint of a user. The cover layer (218) includes a first major surface (228) and a second major surface (230) opposite to the first major surface (228), and is disposed on the LCD panel (204). The cover layer (218) defines a first axis (AA?) normal to the first major surface (228). The optical sensor module (226) receives a light from the predetermined fingerprint sensing area (222) to detect the fingerprint. The optical sensor module (226) at least partially overlaps with the predetermined fingerprint sensing area (222) along the plane of the first major surface (228). The optical sensor module (226) defines a second axis (BB?) normal to a surface of the optical sensor module (226).

Abstract: The present invention relates to the technical field of acetonitrile refining, and in particular, to an improved acetonitrile purification process for an ultrahigh performance liquid chromatography-mass spectrometer. The present invention provides an acetonitrile purification process. A high-purity finished product may be obtained by performing operations of oxidation, rectification adsorption, drying, reflux rectification and filtration on industrial acetonitrile and controlling related parameters such as temperature, flow and the like, continuous production is ensured, a light transmittance of the finished product in ultraviolet rays of 200 to 260 nm is greater than or equal to 95%, water and impurities in the industrial acetonitrile are removed, and the requirements of the ultrahigh performance liquid chromatography-mass spectrometer are met; moreover, by controlling process parameters and equipment.

Abstract: An ultrasound scanning control method applied to an ultrasound scanning device is provided. The ultrasound scanning device includes an execution mechanism. The execution mechanism includes a mechanical arm and a probe. The ultrasound scanning control method includes controlling the mechanical arm to drive the probe to move according to a set scanning trajectory for performing an ultrasound scanning detection on a part to be examined. Once an actual pressure value between the probe and the part to be examined in each control cycle is sensed using the force sensor, a pressure value between the probe and the part to be examined is controlled to be a constant value based on the actual pressure value.

Abstract: A system, method, and computer program product for determining the cause of a battery condition, such as a thermal runaway event, are provided. An example battery safety evaluation system may include a battery having a housing and an interior battery compartment. The battery safety evaluation system may include internal sensing elements attached to the interior battery compartment and configured to capture data representative of an internal battery condition event. The battery safety evaluation system may further include an external sensing element attached to the battery housing and configured to capture data representative of an external battery condition event. In addition, the battery safety evaluation system may include a controller for determining a cause of the battery condition based at least in part on a time sequence of events generated from the internal data captured by the internal sensing element and the external data captured by the external sensing element.

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: A hoist for transferred materials in an underground auxiliary transportation system and a method thereof. The hoist includes a hoist body, guiding plates mounted at two ends of the hoist body through bolts, twist locks mounted at four ends of the hoist body, a driving unit configured to drive twist locks to rotate, transmission assemblies configured to connect the twist locks with the driving unit, ejector pins configured to prevent the twist locks from rotating mistakenly, and a sensor configured to control the hoist to operate and provide protection safety.

Abstract: A powder-based three-dimensional printing (3DP) method, device and system, and a computer-readable storage medium. The method includes: analyzing printing images of layers corresponding to a part to be printed to determine a target print image and adding a preset mark to the target print image which includes a print image of a target layer that causes a previous powder layer of the target layer to displace during printing; acquiring an image to be printed of a current layer to be printed; identifying the image to be printed to determine whether the preset mark exists on the image to be printed; and if yes, processing the current layer to be printed and/or a previous powder layer of the current layer to be printed such that the previous powder layer of the current layer does not move with powder spreading of the current layer.

Abstract: An integrated optical component coupled to a circuit board of an aerosol sensor is provided. The integrated optical component comprising a medium including a first, second plane, third plane, wherein the first plane is adjacent to a detection area, the second plane is positioned about a photosensor, and the third plane is opposite an angle formed by an intersection of the first and second plane. The integrated optical component further comprising a first lens configured on the first sidewall, the first lens configured to receive incident light from the detection area and focus the incident light onto a reflector through the medium, the reflector configured on the third sidewall, the reflector configured to reflect the incident light towards a second lens, and the second lens configured on the second sidewall, the second lens configured to receive the incident light from the reflector and focus the incident light to the photosensor.

Abstract: Various embodiments of the present disclosure are directed towards a semiconductor processing system including an overlay (OVL) shift measurement device. The OVL shift measurement device is configured to determine an OVL shift between a first wafer and a second wafer, where the second wafer overlies the first wafer. A photolithography device is configured to perform one or more photolithography processes on the second wafer. A controller is configured to perform an alignment process on the photolithography device according to the determined OVL shift. The photolithography device performs the one or more photolithography processes based on the OVL shift.