w2v2_base_dutch / README.md
Jakobkee's picture
Update README.md
cecf06b
|
raw
history blame
354 Bytes
Pretrained and finetuned wav2vec2.0 Base model on Flemish data.
Pre-training data: CGN (all components, VL) + VrijeWesten (2000h) + VRT (500h)
Finetuning data: CGN (all components, VL)
Usage:
python
~ import fairseq
~ ckpt = '/path/to/modeldir/checkpoint_best.pt'
~ model, cfg, task = fairseq.checkpoint_utils.load_model_ensemble_and_task([ckpt])