whisper base ar quran This model is a fine tuned version of openai/whisper base on the None dataset. It achieves the following results on the evaluation set: Loss: 0.0839 Wer: 5.7544 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: 0.0001 train batch size: 16 eval batch size: 8 seed: 42 distributed type: multi GPU num devices: 8 total train batch size: 128 total eval batch size: 64 optimizer: Adam with betas=(0.9,0.999) and epsilon=1e 08 lr scheduler type: linear lr scheduler warmup steps: 500 training steps: 5000 mixed precision training: Native AMP Training results Training Loss Epoch Step Validation Loss Wer : : : : : : : : : : 0.1092 0.05 250 0.1969 13.3890 0.0361 0.1 500 0.1583 10.6375 0.0192 0.15 750 0.1109 8.8468 0.0144 0.2 1000 0.1157 7.9754 0.008 0.25 1250 0.1000 7.5360 0.0048 1.03 1500 0.0933 6.8227 0.0113 1.08 1750 0.0955 6.9638 0.0209 1.13 2000 0.0824 6.3586 0.0043 1.18 2250 0.0830 6.3444 0.002 1.23 2500 0.1015 6.3025 0.0013 2.01 2750 0.0863 6.0639 0.0014 2.06…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy