Datasets:
Tasks:
Automatic Speech Recognition
Formats:
parquet
Languages:
Slovak
Size:
100K - 1M
ArXiv:
License:
Download data/train-00039-of-00123.parquet from NaiveNeuron/SloPalSpeech: direct link, hf CLI and curl.
- Browser
- Download file 494 MB
-
https://huggingface.co/datasets/NaiveNeuron/SloPalSpeech/resolve/main/data/train-00039-of-00123.parquet
- Command line
-
hf download hf://datasets/NaiveNeuron/SloPalSpeech/data/train-00039-of-00123.parquet
-
curl -L -o train-00039-of-00123.parquet https://huggingface.co/datasets/NaiveNeuron/SloPalSpeech/resolve/main/data/train-00039-of-00123.parquet
494 MB
- Xet hash:
- 5f01847cda89eb340ff2db0d9c10adffc30f5e5a219ece71349dc2843a326857
- Size of remote file:
- 494 MB
- SHA256:
- d724ef764f40a9ec44af9226c2c910ecd893dd8e76e645c8200677a5c7330665
·
Xet efficiently stores Large Files inside Git, intelligently splitting files into unique chunks and accelerating uploads and downloads. More info.