4 results about "Comparison standard" patented technology

The invention belongs to the technical field of artificial intelligence, and discloses a big language model comprehensive capability assessment method and a related device, and the method comprises the steps: obtaining a multi-difficulty evaluation data set of each capability dimension, dynamically evaluating and extracting three core indexes of critical difficulty, capability balance and stability of a model, and then generating a comprehensive capability assessment result. According to the method, a traditional single accuracy index is abandoned, weak items of the ability are positioned through critical difficulty, 'biases' are prevented through balance, expression fluctuation under the same difficulty is evaluated through stability, and three-dimensional accurate evaluation is achieved. The dynamic difficulty and capability boundary mapping are combined, the evaluation efficiency is remarkably improved, the method is suitable for research and development optimization and model selection comparison of a large language model, capability comparison standards can be provided in the model selection stage, data matching and architecture design can be guided in the research and development stage, short boards are accurately positioned in the evaluation stage, and early warning capability attenuation in the application stage is achieved. Therefore, core technical support is provided for full-life-cycle management of large language models in various industries.

The invention discloses a safety management method and system for an intelligent construction site, and belongs to the field of safety management.The safety management method for the intelligent construction site comprises the following steps that a risk feature database containing various illegal behaviors is constructed and serves as a comparison standard for identifying the illegal behaviors, and the risk feature database is a dynamic database and serves as a comparison standard for identifying the illegal behaviors; compared with the prior art, the method has the beneficial effects that by constructing the dynamic risk database, the accurate digital definition and continuous learning ability of various illegal behaviors are realized, the illegal behaviors can be automatically identified in an all-weather manner by means of real-time information acquisition and intelligent comparison, intervention is immediately performed through positioning voice warning, and the risk of the illegal behaviors can be effectively eliminated. The monitoring instantaneity and coverage range are greatly improved, a complete closed loop from warning sending to behavior correction (or upgrading reporting) is ensured through continuous tracking and state management, each violation event can be effectively tracked and processed, and the risk of safety accidents caused by untimely correction is effectively reduced.

This invention belongs to the field of artificial intelligence technology and discloses a method and related apparatus for comprehensive capability evaluation of large language models. This method acquires multi-difficulty evaluation datasets across various capability dimensions, dynamically evaluates and extracts three core indicators: critical difficulty, capability balance, and stability, and then generates a comprehensive capability evaluation result. It abandons the traditional single accuracy indicator, using critical difficulty to pinpoint weaknesses, balance to prevent "uneven development," and stability to assess performance fluctuations under the same difficulty level, achieving three-dimensional accurate evaluation. By combining dynamic difficulty with capability boundary mapping, it significantly improves evaluation efficiency and is applicable to the R&D optimization and selection comparison of large language models. It can provide capability comparison standards during the selection stage, guide data allocation and architecture design during the R&D stage, accurately identify shortcomings during the evaluation stage, and warn of capability decay during the application stage, thus providing core technical support for the full lifecycle management of large language models in various industries.

The invention belongs to the technical field of data management, and particularly relates to a real estate data non-corresponding auditing method and system based on big data and a terminal. According to the method, on the basis of a comparison data set constructed by data to be detected and a preset database, after real estate data information is matched with initial comparison items in the data set one by one, according to a comparison result of a matching value and a matching threshold value, all the initial comparison items with the high matching degree are screened and collected, and therefore standard data is formed. Through construction of standard data, systematic deviation possibly introduced by taking a single fixed data source as a comparison standard is avoided, the accuracy of subsequent non-corresponding item determination is improved, and a solid foundation is laid for implementation of the whole auditing process.