How to load dataset downloaded with huggingface cli?

[minmie]

I use cli to download mozilla-foundation/common_voice_13_0(only download part of it)：

huggingface-cli download --repo-type dataset --resume-download mozilla-foundation/common_voice_13_0 audio/zh-CN/train/zh-CN_train_0.tar
huggingface-cli download --repo-type dataset --resume-download mozilla-foundation/common_voice_13_0 transcript/zh-CN/train.tsv
huggingface-cli download --repo-type dataset --resume-download mozilla-foundation/common_voice_13_0 languages.py common_voice_13_0.py release_stats.py n_shards.json

And I load the dataset as follow:

from datasets import load_dataset, DatasetDict
DATASET = 'mozilla-foundation/common_voice_13_0' 
common_voice["train"] = load_dataset(DATASET, "zh-CN", split="train[:100]")`

but I found it still redownload from the hub. so, how to load dataset downloaded with huggingface cli?

[wygao8]

As mentioned in issue#6733, this function is not implemented yet.

[yeatscircle]

I also encountered this problem, when I wanted to download allenai/c4, I couldn't select the variants.Is there a solution now?