I faced a trouble when I'm trying to

I faced a trouble when I'm trying to reproduce flashRAG's benchmark with llamaindex, the first step is build index with a corpus about 16GB. But it seems that my memory is not enough to finish index building. Here is my source code, and link of corpus is https://huggingface.co/datasets/ignore/FlashRAG_datasets/resolve/main/retrieval-corpus/wiki18_100w.jsonl?download=true