whisper-medium-dv

This model is a fine-tuned version of openai/whisper-medium on the mozilla-foundation/common_voice_13_0 dv dataset. It achieves the following results on the evaluation set:

• Loss: 0.2998
• Wer: 8.9578

To reproduce this run, execute the command in run.sh. Note that you will require the DeepSpeed package, which can be pip installed with:

pip install --upgrade deepspeed

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 4000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0349	3.58	1000	0.1622	9.9437
0.0046	7.17	2000	0.2288	9.5090
0.0007	10.75	3000	0.2820	9.0952
0.0	14.34	4000	0.2998	8.9578

Framework versions

• Transformers 4.31.0.dev0
• Pytorch 2.0.1+cu118
• Datasets 2.13.1.dev0
• Tokenizers 0.13.3