Prompt word optimization method for processing multi-label classification task for large language model

The invention provides a cue word optimization method for processing a multi-label classification task for a large language model, and belongs to the technical field of artificial intelligence, and the method comprises the steps: determining cue word blocks corresponding to a target multi-label classification task; for each cue word block, generating cue word templates with different expression modes by adopting a large language model, and constructing a multi-dimensional cue word template space; the maximum number of iterations is set, a target cue word template is randomly determined from the multi-dimensional cue word template space in each iteration, and based on the target cue word template, multi-label classification calculation is carried out on a given data set by utilizing a large language model to obtain a marking result; calculating a performance result of the target cue word template based on the marking result and a preset performance index; and judging whether the maximum number of iterations is