Datasets:
Languages:
English
Multilinguality:
monolingual
Size Categories:
1M<n<10M
Language Creators:
found
Annotations Creators:
no-annotation
Source Datasets:
original
Tags:
License:
hello, how to convert these dataset to .bin and .idx to train gpt2?
#8
by
etoilestar
- opened
hello, how to convert these dataset to .bin and .idx to train gpt2?
why convert it to .bin and .idx
I think it is not needed
thanks
thanks for your reply, in fact, I am running Deepspeed-Megatron, a well-known project, which needs me to provide a .json file to preprocess, however, this dataset is totally different from https://mega.nz/#F!EZZD0YwJ!9_PlEQzdMVLaNdKv_ICNVQ!cc4RgQQZ, which is mentioned in https://github.com/microsoft/Megatron-DeepSpeed, could you give me some guidance?