LongLoRA: Efficient Fine-tuning of Long-Context Large Language Models

arXiv:2309.12307v3 Announce Type: replace-cross
Abstract: We present LongLoRA, an efficient fine-tuning approach that extends the context sizes of pre-trained large language models (LLMs), with limited computation cost. Typically, training LLMs with long context sizes is computationally expensive, requiring extensive training hours and GPU resources. For example, training on the context length of 8192 needs 16x computational costs in self-attention layers as that of 2048. In this paper, we speed up the context extension of LLMs in two aspects. On the …

abstract arxiv computation context cost cs.ai cs.cl cs.lg example fine-tuning gpu gpu resources language language models large language large language models llms resources training training llms type