Word Recovery in Large Language Models Enables Character-Level Tokenization Robustness 文章

ArXiv CS.CL2026-08-05PAPERen作者: Zhipeng Yang, Shu Yang, Lijie Hu, Di Wang
大语言模型
查看原文 →
关系图谱

Word Recovery in Large Language Models Enables Character-Level Tokenization Robustness · 相关技术

相关技术

CLEARDDECross-lingual biomedical entity linkingContrastive ReflectionCoreCondition Dropout (ConD)Composed Object RetrievalCamouflaged object detectionCalibrated Entropy ScoreCTECRISCOVERCOVCODICANONBotAuxiliary Regression LossAutoregressive large language modelsAudio Mamba EncoderArtificial Language LearningAdaptive Batch ScalingAction UnitsATEARGANOANNCodingAHATABLEABI@State3D Segment Anything ModelACEAbstractACT