2411.09009v2
arXiv License
在大詞彙語言模型中削減你的損失
Cut Your Losses in Large-Vocabulary Language Models
Erik Wijmans、Brody Huval、Alexander Hertzberg、Vladlen Koltun、Philipp Krähenbühl（Apple）
Cross-EntropyMemory EfficiencyLarge VocabularyLLM TrainingCUDA KernelFlash MemoryApple
Liang-Hsun Huang 許願