Roberta model called SRoBERTa trained on WOL dataset (OSCAR + Leipzig + srWac) for Serbian language. Attention heads distilled 6, batch size 64, group size 64, epochs 2, test split 0.05(~1mil groups)
4e75a39 - Xet hash:
- 47158ed370f527f49b7ca3a37ecb2da610defab8f024c761457aacdf0ca94cda
- Size of remote file:
- 322 MB
- SHA256:
- e7d6251e1f8cc0a6dc6b494e952a1413792f71ce3e78a211a26d883c6cbaa0c5
·
Xet efficiently stores Large Files inside Git, intelligently splitting files into unique chunks and accelerating uploads and downloads. More info.