Abstract: Systems and methods are disclosed for detecting hallucinations in large language models (LLMs). An example method includes receiving a first prompt for submission to the first LLM, generating, using the first LLM, a plurality of semantically equivalent prompts to the first prompt, generating, using the first LLM, a first response to the first prompt and a plurality of second responses to the plurality of semantically equivalent prompts, generating, using a second LLM, a plurality of third responses to the semantically equivalent prompts, generating a semantic consistency score for the first response based at least in part on the first prompt, the plurality of semantically equivalent prompts, the plurality of second responses, and the plurality of third responses, and determining whether or not the first response is an accurate response to the first prompt based at least in part on the semantic consistency score.

Abstract: The present disclosure provides an amplification primer for sequencing library construction comprising a primer sequence fragment complementary to a target fragment and a base G ligated to the 5? end of the primer sequence fragment, wherein the amplification primer and the target fragment are not complementary at the base G. Further disclosed are a method for constructing a sequencing library, a sequencing library, a kit for constructing a sequencing library, a mutant site detection kit, a chromosome ploidy detection kit, a gene fusion detection kit, and a sequencing method.

Abstract: The computing-in-memory circuit includes: an SRAM memory cell array including at least one memory cell connected between a first bit line and a second bit line, the memory cell includes a first inverter and a second inverter cross-coupled with each other, and the first inverter and the second inverter have an asymmetric configuration with respect to each other; a control circuit configured to: receive a first input signal, a second input signal and an operation mode control signal, process the first input signal and the second input signal according to operation mode control signal, so as to obtain a processed first input signal and a processed second input signal, and apply the processed first input signal and the processed second input signal to the first bit line and the second bit line, respectively; and a readout circuit configured to read out data stored in memory cell from the memory cell.

Abstract: Systems and methods are provided for improved point querying of a database. The index values are separated from data and retained in cache memory to allow access without requiring a disk input/output (I/O) operation and thereby having less latency resulting from such disk I/O operations. The index values can be compressed using an algorithm such as Crit-Bit-Trie to allow storage of the index values in limited cache memory space. The index values can be selected for storage according to a least recently used approach when cache memory is insufficient to store all index values to maintain a hit rate for the cached portion and reduce the disk I/O operations.

Abstract: Augmented Denoising Diffusion Implicit Models (“DDIMs”) using a latent trajectory optimization process can be used for image generation and manipulation using text input and one or more source images to create an output image. Noise bias and textual bias inherent in the model representing the image and text input is corrected by correcting trajectories previously determined by the model at each step of a diffusion inversion process by iterating multiple starts the trajectories to find determine augmented trajectories that minimizes loss at each step. The trajectories can be used to determine an augmented noise vector, enabling use of an augmented DDIM and resulting in more accurate, stable, and responsive text-based image manipulation.

Abstract: Provided is a method for preparing copper azide and cuprous azide encapsulated by conductive metal-organic framework. The method uses a conductive copper-containing metal-organic framework material as a precursor, and completes the azidation of the precursor by means of a liquid-solid electrochemical azidation reaction. Copper azide and cuprous azide nanocrystals are highly uniformly embedded within a conductive framework, which may effectively avoid the agglomeration of copper azide and cuprous azide, and reduce static charge generated by friction, displacement, and the like. Meanwhile, the conductive framework may promote the effective transfer of charge, avoid the accumulation of static charge, and improve the electrostatic safety.

Abstract: A glass part includes a substrate and a texture reinforcement layer disposed on at least one surface of the substrate. The texture reinforcement layer includes a plurality of reinforcement units. Cross sections of an outer surface of each reinforcement unit in at least a horizontal direction and a vertical direction are in a hyperbolic shape or a parabolic shape.

Abstract: A semiconductor device and a method for manufacturing the same. The method comprise: forming a first field-effect transistor (FET) disposed on a substrate and a first isolation layer disposed on the first FET; forming a first through hole in the first isolation layer, where a metal layer is deposited in the first through hole and is electrically connected to the first FET; forming a second isolation layer, which is disposed on the first isolation layer and the metal layer; and forming a second FET which is disposed on the second isolation layer, where a second through hole is disposed in the second FET and the second isolation layer, a metal material filled in the second through hole serves as a first contact plug, and the first contact plug is electrically connected to the metal layer. The metal layer serves as a power distribution network for both FETs.

Abstract: A method including receiving a user input from a user device. The method also includes generating test inputs including the user input and modified inputs. The user input is processed with a rephrasing model to form the modified inputs. The method also includes executing a test model to generate test outputs, including an original test output and modified test outputs, from processing the test inputs. The method also includes generating similarity scores by performing similarity comparisons among the test outputs. The method also includes determining a model confidence from the similarity scores. The method also includes routing the user input responsive to the model confidence satisfying or failing to satisfy a confidence threshold.

Abstract: Some embodiments relate to a photothermal-curing resin composition, and a preparation method therefor and the use thereof. The resin composition includes a combination of specific parts of a prepolymer, a monomer, a photoinitiator, a thermal initiator, fumed silica and a glass powder. By means of adding specific parts of the photoinitiator in conjunction with the thermal initiator, the completion rate of a curing reaction is increased, the resin composition is more thoroughly cured, and the residual monomer rate after curing is lower, and specific parts of the fumed silica and barium-containing glass powder are further added as fillers, such that the hardness, strength and wear resistance of the resin composition are effectively enhanced; therefore, a product resulting from the 3D printing of the finally obtained resin composition completely meets the requirements for being used as temporary crowns and bridges.

Abstract: Aspects of the present disclosure provide techniques for active multifidelity machine learning. Embodiments include selecting, based on one or more criteria, a first subset of unlabeled training data for manual review and a second subset of unlabeled training data for providing to a pre-trained machine learning model for automated labeling. Embodiments include receiving manual label data for the first subset of unlabeled training data. Embodiments include providing inputs to the pre-trained machine learning model based on a subset of the manual label data and the second subset of training data. Embodiments include receiving, as outputs from the pre-trained machine learning model, automated label data for the second subset of unlabeled training data. Embodiments include generating a training data set for a target machine learning model based on the set of unlabeled training data, the manual label data, and the automated label data.

Abstract: A method includes receiving a user input including natural language text. The method also includes generating modified inputs from the user input. The method also includes executing a machine learning model on the modified inputs to generate model outputs. The method also includes sampling the model outputs using a statistical sampling strategy to generate sampled model outputs. The method also includes clustering the sampled model outputs into clusters. The method also includes generating a confidence metric of the clusters. The method also includes routing, automatically in a computing system, the user input based on whether the confidence metric satisfies a threshold value.

Abstract: Augmented Denoising Diffusion Implicit Models (“DDIMs”) using a latent trajectory optimization process can be used for image generation and manipulation using text input and one or more source images to create an output image. Noise bias and textual bias inherent in the model representing the image and text input is corrected by correcting trajectories previously determined by the model at each step of a diffusion inversion process by iterating multiple starts the trajectories to find determine augmented trajectories that minimizes loss at each step. The trajectories can be used to determine an augmented noise vector, enabling use of an augmented DDIM and resulting in more accurate, stable, and responsive text-based image manipulation.

Abstract: A Low Earth orbit (LEO) satellite congestion control routing method is disclosed, including: Step 1: A LEO satellite periodically detecting queue lengths of a plurality of ports and informing a Geostationary Earth Orbit (GEO) satellite of queue lengths of congested ports in the plurality of ports; Step 2: The GEO satellite sending the queue lengths of congested ports to a ground station, and a computing center of the ground station calculating link status weights according to the queue lengths of congested ports; Step 3: According to the link status weights, the computing center of the ground station determining a congestion area; Step 4: The computing center of the ground station calculating a routing table of satellites in the congestion area, and the ground station sending the routing table to a LEO satellite inside the congestion area; Step 5: The LEO satellite in the congestion area receiving the routing table, and performing end-to-end transmission of data packets to a destination LEO satellite accordin

Abstract: A glass substrate and a method for preparing the same, and an electronic device are provided. Two opposite sides of the glass substrate respectively have a first surface compressive stress layer and a second surface compressive stress layer. A surface compressive stress of the first surface compressive stress layer is greater than a surface compressive stress of the second surface compressive stress layer, and a depth of the first surface compressive stress layer is less than a depth of the second surface compressive stress layer.

Abstract: Systems and methods are disclosed for model based document image enhancement. Instead of requiring paired dirty and clean images for training a model to clean document images (which may cause privacy concerns), two models are trained on the unpaired images such that only the dirty images are accessed or only the clean images are accessed at one time. One model is a first implicit model to translate the dirty images from a source space to a latent space, and the other model is a second implicit model to translate the images from the latent space to clean images in a target space. The second implicit model is trained based on translating electronic document images in the target space to the latent space. In some implementations, the implicit models are diffusion models, such as denoising diffusion implicit models based on solving ordinary differential equations.

Abstract: Disambiguity in large language models (LLMs) includes receiving an original query in a user interface, generating an ambiguity query from the original query, and sending, via an application programming interface (API) of an LLM, the ambiguity query to the LLM. The ambiguity query includes the original query and training the LLM to recognize ambiguities. The method further includes receiving, via the API and responsive to the ambiguity query, a binary response and detecting, based at least in part on the binary response, the original query as ambiguous. Disambiguity may include detecting an ambiguity location in the original query using perturbed queries and the LLM.

Abstract: The present disclosure provides a ceramic material for improving surface adhesion of dental zirconia, a restorative material and a preparation method and a adhesion method thereof. The ceramic material includes SiO2, B2O3, GeO2, Al2O3, Li2O, Nb2O5, Na2O, ZrO2, P2O5, and a rare earth oxide. In the present disclosure, the ceramic material includes GeO2.

Abstract: The present disclosure discloses a lithium disilicate glass-ceramic with high strength and high transparency and a preparation method and use thereof. A raw material composition of the lithium disilicate glass-ceramic comprises: 63-75 wt % of SiO2, 13-18 wt % of Li2O, 1-6 wt % of Al2O3, 1-10 wt % of K2O, 2-6 wt % of P2O5, 0-4 wt % of an additive and 0-10 wt % of a colorant; a main crystal phase of the lithium disilicate glass-ceramic is lithium disilicate crystals, and an impurity phase of the lithium disilicate glass-ceramic is any one or a combination of at least two selected from the group consisting of lithium metasilicate, lithium phosphate and quartz; the lithium disilicate crystal has a size larger than 700 nm and a length-diameter ratio not less than 3.

Abstract: Provided are a fluorescent glass ceramic with high transparency and a preparation method and use thereof. The fluorescent glass ceramic includes the following raw materials by mass percentage: 63 wt % to 70 wt % of SiO2, 13 wt % to 16 wt % of Li2O, 1 wt % to 6 wt % of Al2O3, 1 wt % to 10 wt % of K2O, 2 wt % to 6 wt % of P2O5, 0.5 wt % to 3.5 wt % of CeO2, 0 wt % to 4 wt % of an additive, 1 wt % to 4 wt % of a lanthanide oxide with an atomic number of 59 to 71, and 0 wt % to 8 wt % of a colorant. The fluorescent glass ceramic has a lithium metasilicate crystal as a principal crystalline phase, and the lithium metasilicate crystal has a layered or plate-like structure and a grain size of 0.1 ?m to 1.5 ?m.