hiyouga/LLaMA-Factory

数据集过大导致加载数据集时内存爆掉,似乎在哪里看到可以直接加载tokenize之后的数据进行训练

Closed this issue · 1 comments

Reminder

  • I have read the README and searched the existing issues.

Reproduction

想要直接加载tokenized 之后的文本进行训练,但是招不到具体使用方法写在哪里了,请问有具体的地址吗

Expected behavior

No response

System Info

No response

Others

No response