wav2vec2 child en tokenizer 4 This model is a fine tuned version of facebook/wav2vec2 xls r 300m on the None dataset. It achieves the following results on the evaluation set: Loss: 1.4709 Wer: 0.3769 Model description More information needed Intended uses & limitations More information needed Training and evaluation data More information needed Training procedure Training hyperparameters The following hyperparameters were used during training: learning rate: 0.0003 train batch size: 24 eval batch size: 24 seed: 42 gradient accumulation steps: 2 total train batch size: 48 optimizer: Adam with betas=(0.9,0.999) and epsilon=1e 08 lr scheduler type: linear lr scheduler warmup steps: 500 num epochs: 30 mixed precision training: Native AMP Training results Training Loss Epoch Step Validation Loss Wer : : : : : : : : : : 0.0334 1.72 100 1.4709 0.3769 0.0332 3.45 200 1.4709 0.3769 0.0343 5.17 300 1.4709 0.3769 0.032 6.9 400 1.4709 0.3769 0.0332 8.62 500 1.4709 0.3769 0.0327 10.34 600 1.4709 0.3769 0.0331 12.07 700 1.4709 0.3769 0.0334 13.79 800 1.4709 0.3769 0.0319 15.52 900 1.4709 0.3769 0.0338 17.24 1000 1.4709 0.3769 0.0321 18.97 1100 1.4709 0.3769 0.0367 20.69 1200 1.4709 0.3769 0.0331…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy