Datasets:
Tasks:
Automatic Speech Recognition
Formats:
parquet
Languages:
Slovak
Size:
100K - 1M
ArXiv:
License:
Download data/train-00047-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-00047-of-00123.parquet
- Command line
-
hf download hf://datasets/NaiveNeuron/SloPalSpeech/data/train-00047-of-00123.parquet
-
curl -L -o train-00047-of-00123.parquet https://huggingface.co/datasets/NaiveNeuron/SloPalSpeech/resolve/main/data/train-00047-of-00123.parquet
493 MB
- Xet hash:
- ad2762ec3bb87f4c4885d6fb49c6b97d9d487c2398858f37ea279b34e6cfe444
- Size of remote file:
- 493 MB
- SHA256:
- ae3fa26554e76d5a5a34747d486ecb70ba82b921e2e1e77932c79156a09351c0
·
Xet efficiently stores Large Files inside Git, intelligently splitting files into unique chunks and accelerating uploads and downloads. More info.