How to finetune w2v_large_lv_fsh_swbd_cv_ftsb300_updated.pt on custom dataset

[shahzebali42]

Fairseq has updated wav2vec2 models on their website.

I want to fine tune w2v_large_lv_fsh_swbd_cv_ftsb300_updated.pt model on my custom dataset. by following your git i have fine tuned w2v_large_lv_fsh_swbd_cv.pt on my custom dataset but when i start to fine tune w2v_large_lv_fsh_swbd_cv_ftsb300_updated model weights it gives me a config error.

Can you please guide me what can i do to finetune model using w2v_large_lv_fsh_swbd_cv_ftsb300_updated as pretrained model.

error :

Key 'target_dict' not in 'AudioPretrainingConfig' full_key: target_dict reference_type=Optional[AudioPretrainingConfig] object_type=AudioPretrainingConfig