Datasets:
Tasks:
Automatic Speech Recognition
Formats:
parquet
Languages:
Slovak
Size:
100K - 1M
ArXiv:
License:
Download data/train-00001-of-00123.parquet from NaiveNeuron/SloPalSpeech: direct link, hf CLI and curl.
- Browser
- Download file 493 MB
-
https://huggingface.co/datasets/NaiveNeuron/SloPalSpeech/resolve/main/data/train-00001-of-00123.parquet
- Command line
-
hf download hf://datasets/NaiveNeuron/SloPalSpeech/data/train-00001-of-00123.parquet
-
curl -L -o train-00001-of-00123.parquet https://huggingface.co/datasets/NaiveNeuron/SloPalSpeech/resolve/main/data/train-00001-of-00123.parquet
493 MB
- Xet hash:
- 5e37080933f7d1730ecb95cb00c7582aff5aad68ac8bf095b4c888f758ea7ab3
- Size of remote file:
- 493 MB
- SHA256:
- ec29556f745f78a29f6010adf3521d4bc8e77fc657324760c98e154429d0817b
·
Xet efficiently stores Large Files inside Git, intelligently splitting files into unique chunks and accelerating uploads and downloads. More info.