Aiming at the problem of reverse-design of mechanism,a method based on the matching of trajectory code-chains is presented.The motion trajectory of mechanism is described with code-chain,which is normalized to simplif...Aiming at the problem of reverse-design of mechanism,a method based on the matching of trajectory code-chains is presented.The motion trajectory of mechanism is described with code-chain,which is normalized to simplify the operation of geometric transformation.The geometric transforma-tion formulas of scale,mirror and rotation for trajectory code-chain are defined,and the reverse de-sign for mechanism trajectory is realized through the analysis and solution of similarity matching between the desired trajectory and the predefined trajectory.The algorithm program and prototype system of reverse design for mechanism trajectory are developed.Application samples show that the method can break the restriction of trajectory patterns in matching,meet the demand of partial match-ing,and overcome the influence of geometric transformation of trajectory on the reverse design for mechanism.展开更多
大语言模型因其出色的理解和生成能力被广泛应用于自动化智能助手的开发。然而,它们在处理复杂问题时,常因训练数据庞杂等局限性而难以调用正确的工具和生成准确的函数名称及参数。为了提高在对复杂问题进行任务规划时模型生成工具调用...大语言模型因其出色的理解和生成能力被广泛应用于自动化智能助手的开发。然而,它们在处理复杂问题时,常因训练数据庞杂等局限性而难以调用正确的工具和生成准确的函数名称及参数。为了提高在对复杂问题进行任务规划时模型生成工具调用的准确性,提出一种基于思维链的提示方法—思维代码(Reasoning to Annotation and Coding,ReACo),充分利用预训练的数据,通过代码和注释结合的任务规划提示方式增强语言模型对复杂任务的理解能力,并基于此提出一种新的大语言模型思维提示框架ReACoGPT。基于ReACoGPT提示的语言模型能够准确调用多个插件,依据事实提供富有逻辑的任务规划能力,从而在保持任务规划逻辑性的同时准确使用需求的真实数据。实验结果表明,相较于现有方法,ReACoGPT在RestBench数据集上的正确率、正确路径率及解决长度等多项指标得到提高,证实了ReACo提示方式能够增强大语言模型的规划和推理能力,有效利用大量训练数据对任务进行有效规划,促进了大语言模型在工具学习方面的进一步发展。展开更多
基金supported by National Hi-tech Research and Development Program of China(863 Program,No.2006AA04Z114)Research Fund for the Doctoral Program of Higher Education,China(No.20040335060)Provincial Scientific Personnel Educational Foundation of Zhejiang,China(No.R603240).
文摘Aiming at the problem of reverse-design of mechanism,a method based on the matching of trajectory code-chains is presented.The motion trajectory of mechanism is described with code-chain,which is normalized to simplify the operation of geometric transformation.The geometric transforma-tion formulas of scale,mirror and rotation for trajectory code-chain are defined,and the reverse de-sign for mechanism trajectory is realized through the analysis and solution of similarity matching between the desired trajectory and the predefined trajectory.The algorithm program and prototype system of reverse design for mechanism trajectory are developed.Application samples show that the method can break the restriction of trajectory patterns in matching,meet the demand of partial match-ing,and overcome the influence of geometric transformation of trajectory on the reverse design for mechanism.
文摘大语言模型因其出色的理解和生成能力被广泛应用于自动化智能助手的开发。然而,它们在处理复杂问题时,常因训练数据庞杂等局限性而难以调用正确的工具和生成准确的函数名称及参数。为了提高在对复杂问题进行任务规划时模型生成工具调用的准确性,提出一种基于思维链的提示方法—思维代码(Reasoning to Annotation and Coding,ReACo),充分利用预训练的数据,通过代码和注释结合的任务规划提示方式增强语言模型对复杂任务的理解能力,并基于此提出一种新的大语言模型思维提示框架ReACoGPT。基于ReACoGPT提示的语言模型能够准确调用多个插件,依据事实提供富有逻辑的任务规划能力,从而在保持任务规划逻辑性的同时准确使用需求的真实数据。实验结果表明,相较于现有方法,ReACoGPT在RestBench数据集上的正确率、正确路径率及解决长度等多项指标得到提高,证实了ReACo提示方式能够增强大语言模型的规划和推理能力,有效利用大量训练数据对任务进行有效规划,促进了大语言模型在工具学习方面的进一步发展。
