Abstract: Disclosed is a reactive power control method for suppressing a transient overvoltage in a DC sending-end power grid of new energy. The method estimates the trend of grid voltage changes by using current and historical signals of the grid voltage, calculates the reactive power command offset value to achieve the effect of reactive power lead control, and combines the optimized reactive power value after lag compensation. The superposition of the lag compensation and the reactive power optimization generates a final command value for a new energy power source, and this command value can ensure that new energy effectively outputs reactive power to participate in the suppression of the transient overvoltage in a DC converter station.

Abstract: Methods, apparatuses, and computer-program products are disclosed. The method may include storing, on a local machine, a first vector database and a second vector database that include first and second vector embeddings of first and second context information. The method may include receiving a query requesting generation of code. The method may include comparing a third vector embedding that is based on the query with the plurality of first vector embeddings and the plurality of second vector embeddings to identify one or more first vector embeddings and one or more second vector embeddings. The method may include generating a prompt that may include the query, first information and second information corresponding with the one or more first and second vector embeddings. The method may include generating, with a generative artificial intelligence (AI) model, a response to the query based on the query, the first information, and the second information.

Abstract: Systems and techniques are described for providing a power supply glitch detector. For example, a computing device can generate (e.g., using a first supply scale-down engine connected to a power supply) based on a lower voltage threshold, a first voltage close to a reference voltage, compare (e.g., using a first comparator) the first voltage with the reference voltage, and generate, based on determining the first voltage is less than the reference voltage, a low flag indicating detection of a fault in the power supply. The computing device can generate (e.g., using a second supply scale-down engine connected to the power supply), based on an upper voltage threshold, a second voltage close to the reference voltage, compare (e.g., using a second comparator) the second voltage with the reference voltage, and generate, based on determining the second voltage is greater than the reference voltage, a high flag indicating detection of the fault.

Abstract: Disclosed is a flexible hybrid electronic system for intraoperative intracranial monitoring. The system includes a flexible front-end signal collection circuit, a back-end signal processing circuit and a computer system, wherein the front-end signal collection circuit includes a flexible substrate, an electroencephalogram layer, a flexible sensor array, a temperature-controlled hydrogel interface layer, and an electrothermal film. The flexible sensor array is used for collecting pressure, temperature, electroencephalographic and cerebral tissue oxygen saturation signals, and the electrothermal film regulating the adhesion and peeling of the temperature-controlled hydrogel interface layer and cerebral tissue. The signal processing circuit is electrically connected to the signal collection circuit, so as to receive signals collected by the flexible sensor array and the electroencephalographic layer, analyze then transmit to the computer system.

Abstract: Methods, apparatuses, and computer-program products are disclosed. The method may include storing, on a local machine, a first vector database and a second vector database that include first and second vector embeddings of first and second context information. The method may include receiving a query requesting generation of code. The method may include comparing a third vector embedding that is based on the query with the plurality of first vector embeddings and the plurality of second vector embeddings to identify one or more first vector embeddings and one or more second vector embeddings. The method may include generating a prompt that may include the query, first information and second information corresponding with the one or more first and second vector embeddings. The method may include generating, with a generative artificial intelligence (AI) model, a response to the query based on the query, the first information, and the second information.

Abstract: Embodiments described herein provide a multi-stage rating and re-ranking pipeline for selecting SWE agents for an input issue description. Specifically, a meta-policy may be selected among available agent policies corresponding to a pool of available SWE agents which maximizes the cumulative reward along the trajectory of states (such as status of a file) and actions taken at a series of time steps, and a context of relevant repository information and issue descriptions. By dynamically choosing the most suitable agent policy for each context, the selection pipeline maximizes the expected cumulative reward across all possible contexts. In this way, software issue resolve rate is improved.

Abstract: Embodiments described herein provide A method of fine-tuning a neural network based model. In some embodiments, a system receives, via a data interface, a training dataset including a plurality of input samples. The system generates, via a pre-trained neural network based model, a first response based on a first input sample of the plurality of input samples, and a second response based on the first input sample. The system generates, via a trained reward model, a first reward score based on the first input sample and the first response, and a second reward score based on the first input sample and the second response. The system computes a loss function based on the first prompt, the first response, the second response, the first reward score, and the second reward score. The system updates parameters of the neural network based model based on the loss function.

Abstract: Embodiments described herein provide a prompt-based transfer learning method that employs shared latent space prompt tuning). Specifically, a shared latent space is assumed, among all source and target tasks, where each vector in the space captures a basis skill to do a particular task. Given an instance (from either a source task or a target task), it is first encoded into an instance representation vector and then queries the latent space, which yields a skill vector for this instance. This vector modulates a frozen model, via soft prompts which are a simple prompt transformation (the prompt generator in FIG. 3) of the basis skill vector, to generate an answer for the instance. The latent space and prompt transformation are learned end-to-end in upstream pre-training on source tasks.

Abstract: An apparatus, including: a switched capacitor configured to generate a switched capacitor voltage based on an input clock signal and a current; a current digital-to-analog converter (DAC) configured to generate the current based on a first digital signal; a first reference voltage generator configured to generate a first reference voltage; and a first voltage comparing device configured to generate a first frequency deviation detection signal based on a comparison of the switched capacitor voltage to the first reference voltage.

Abstract: The present disclosure discloses a thermophilic recombinant type II pullulanase and the application thereof, and belongs to the technical field of genetic engineering. The present disclosure obtains a thermophilic recombinant type II pullulanase by heterologously expressing type II pullulanase in Escherichia coli. Its optimum pH is 6.6, it has better pH tolerance under the conditions of pH 5.8-8.0, and its optimum temperature is 95° C. After incubating at 95° C. for 10 h, the remaining enzyme activity is greater than 50%. It can exhibit higher specific enzyme activity under strong reducing conditions. For example, adding DTT to the culture environment can increase the specific enzyme activity of Sumo-PulPy by 237.2%. The present disclosure also provides the combined truncation mutant ?28N+?791C of type II pullulanase Sumo-PulPy. The specific enzyme activity of the enzyme mutant is 32.18±0.92 U/mg, which is 5.

Abstract: The present invention relates to use of a benzisoselenazole compound in the preparation of a drug for the treatment of spinaglioma. By applying the benzisoselenazole compound, cell proliferation of the spinaglioma can be effectively inhibited at the cellular level, with an obvious dose dependence. The tumor growth of a tumor-bearing mouse can be effectively inhibited at the animal model level, and the survival period of the tumor-bearing mouse can be prolonged.

Abstract: A preparation method of a polytetrafluoroethylene (PTFE)-based membrane for preventing and removing ices covering wind turbine blades is provided and the method comprises: preparing a membrane into a PTFE rod material with polymerized monomers by using monomer polymerization methods such as blending, pre-compressing and pushing; making the membrane into a PTFE-based homogeneous membrane with micropores and nano and micron scale concave-convex geometrical ultra-structure morphologies under the condition that the membrane is cracked to generate a laminar exfoliated fabric-like structure in the hot calendaring process of the PTFE rod material by using a hot calendaring and fusion polymerization method; and applying the PTFE-based homogeneous membrane to blades of a large wind turbine in operation.

Abstract: Embodiments described herein provide a program synthesis framework that generates code programs through a multi-turn conversation between a user and a system. Specifically, the description to solve a target problem is factorized into multiple steps, each of which includes a description in natural language (prompt) to be input into the generation model as a user utterance. The model in turn synthesizes functionally correct subprograms following the current user utterance and considering descriptions and synthesized subprograms at previous steps. The subprograms generated at the multiple steps are then combined to form an output of program in response to the target problem.

Abstract: The present disclosure provides a direct current (DC) transformer error detection apparatus for a pulsating harmonic signal, including a DC and pulsating harmonic current output module and an external detected input module, where the DC and pulsating harmonic current output module outputs a DC and a DC superimposed pulsating harmonic current to an internal sampling circuit and a self-calibrated standard resistor array; and the internal sampling circuit converts the input DC and the input DC superimposed pulsating harmonic current into a voltage signal, and sends the voltage signal to an analog-to-digital (AD) sampling and measurement component through a front-end conditioning circuit and a detected input channel. The DC transformer error detection apparatus can complete self-calibration for measurement of the DC and the pulsating harmonic signal on a test site.

Abstract: Disclosed herein are pharmaceutical compositions comprising biosynthetic allosteric mTOR inhibitors that can have improved pharmacology and reduced toxicity. Also disclosed herein are methods of treating a condition or disease by administering biosynthetic allosteric mTOR inhibitors.

Abstract: An apparatus, including: a switched capacitor configured to generate a switched capacitor voltage based on an input clock signal and a current; a current digital-to-analog converter (DAC) configured to generate the current based on a first digital signal; a first reference voltage generator configured to generate a first reference voltage; and a first voltage comparing device configured to generate a first frequency deviation detection signal based on a comparison of the switched capacitor voltage to the first reference voltage.

Abstract: Embodiments described herein provide a program synthesis framework that generates code programs through a multi-turn conversation between a user and a system. Specifically, the description to solve a target problem is factorized into multiple steps, each of which includes a description in natural language (prompt) to be input into the generation model as a user utterance. The model in turn synthesizes functionally correct subprograms following the current user utterance and considering descriptions and synthesized subprograms at previous steps. The subprograms generated at the multiple steps are then combined to form an output of program in response to the target problem.

Abstract: The present invention provides a preparation method of a gallium nitride single crystal based on a ScAlMgO4 substrate, comprising following steps: (1) providing a ScAlMgO4 substrate; (2) growing a buffer layer on a surface of the ScAlMgO4 substrate; (3) annealing the buffer layer; (4) growing a GaN crystal on the buffer layer; (5) performing cooling, so that the GaN crystal is automatically peeled off from the ScAlMgO4 substrate. The present invention does not need to use a complex MOCVD process for GaN deposition and preprocessing to make a mask or a separation layer, which effectively reduces production costs; compared with traditional substrates such as sapphire, it has higher quality and a larger radius of curvature, and will not cause a problem of OFFCUT non-uniformity for growing GaN over 4 inches; finally, the present invention can realize continuous growth into a crystal bar with a thickness of more than 5 mm, which further reduces the costs.

Abstract: A concrete interface agent relates to the technical field of concrete surface protection, an ingredient of the concrete interface agent comprises 55 to 100 parts by weight of a nano-calcium salt solution and a nano-SiO2 precursor, 0.1 to 0.4 parts by weight of a surfactant, 30 to 60 parts by weight of a silane coupling agent and 10 to 40 parts by weight of a polydimethylsilane, an ingredient of the nano-calcium salt solution comprises 2 to 5 parts by weight of a calcium hydroxide, 2 to 5 parts by weight of an acid catalyst and 200 to 500 parts by weight of an alcohol-based organic solvent, which can form a coating layer with higher hydrophobic angle on the concrete surface, reduce the water absorption of the concrete, and is not easy to crack after drying, which has more protective effect and longer service life than the existing TEOS interface agent.