2results about How to "Achieve utilization" patented technology

This invention discloses a data annotation method, apparatus, storage medium, and program product, relating to the field of artificial intelligence. The method includes: inputting a target sample into a target language model to obtain N candidate labels, wherein the target sample includes data samples to be labeled; calculating the similarity between each candidate label and S standard labels to obtain N similarity sets, wherein each similarity set includes the similarity between one candidate label and the S standard labels, and the standard labels include labels defined based on domain knowledge and business requirements of the target sample's domain; and determining the target label for the target sample from the S standard labels based on the N similarity sets. This invention solves the technical problem in related technologies where labels directly annotated by models are difficult to utilize directly, leading to low utilization rates of sample labels.

The present application belongs to the technical field of smelting reduction ironmaking, and particularly relates to a smelting method of a mixture of multiple special iron ores. The method comprises the following steps: S1. mixing sea sand ore, high-phosphorus ore and red mud into a mixture according to a proportion; the mixing proportion is: sea sand ore (0-50%), high-phosphorus ore (0-50%): red mud (0-50%); S2. preheating the mixture; S3. feeding the preheated mixture into a smelting furnace for ironmaking. The innovation of the present application lies in solving the problem of large use of special iron ores, improving the use amount of special iron ores, realizing the utilization rate of resources, and achieving good economic benefits (the benefit per ton of iron can reach 526.73 yuan).