wav2vec2 large xls r 300m bg d2 This model is a fine tuned version of facebook/wav2vec2 xls r 300m on the MOZILLA FOUNDATION/COMMON VOICE 8 0 BG dataset. It achieves the following results on the evaluation set: Loss: 0.3421 Wer: 0.2860 Evaluation Commands 1. To evaluate on mozilla foundation/common voice 8 0 with test split python eval.py model id DrishtiSharma/wav2vec2 large xls r 300m bg d2 dataset mozilla foundation/common voice 8 0 config bg split test log outputs 2. To evaluate on speech recognition community v2/dev data python eval.py model id DrishtiSharma/wav2vec2 large xls r 300m bg d2 dataset speech recognition community v2/dev data config bg split validation chunk length s 10 stride length s 1 Training hyperparameters The following hyperparameters were used during training: learning rate: 0.00025 train batch size: 16 eval batch size: 8 seed: 42 gradient accumulation steps: 2 total train batch size: 32 optimizer: Adam with betas=(0.9,0.999) and epsilon=1e 08 lr scheduler type: linear lr scheduler warmup steps: 700 num epochs: 35 mixed precision training: Native AMP Training results Training Loss Epoch Step Validation Loss Wer : : : : : : : : : : 6.8791 1.74 200 3.1902 1.0…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy