Datasets:
Tasks:
Automatic Speech Recognition
Formats:
parquet
Languages:
Slovak
Size:
100K - 1M
ArXiv:
License:
Download data/train-00026-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-00026-of-00123.parquet
- Command line
-
hf download hf://datasets/NaiveNeuron/SloPalSpeech/data/train-00026-of-00123.parquet
-
curl -L -o train-00026-of-00123.parquet https://huggingface.co/datasets/NaiveNeuron/SloPalSpeech/resolve/main/data/train-00026-of-00123.parquet
493 MB
- Xet hash:
- 88cb2be6fa55485e560519b212e5fa2cff69553fefb0cf5d6b2c3644021bbc17
- Size of remote file:
- 493 MB
- SHA256:
- 2f2f82a71da9e2adbaa4e18b860c24c038fd1aec2f8eca89ac69e584901452bd
·
Xet efficiently stores Large Files inside Git, intelligently splitting files into unique chunks and accelerating uploads and downloads. More info.