whisper-base-khmer-aug-v7

This model is a fine-tuned version of openai/whisper-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2500
• Wer: 100.0829

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: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.526	1.0	984	0.2734	104.3671
0.2156	2.0	1968	0.2333	116.4179
0.1647	3.0	2952	0.2246	97.5124
0.1378	4.0	3936	0.2143	95.9646
0.1185	5.0	4920	0.2176	98.6733
0.103	6.0	5904	0.2248	98.3693
0.0905	7.0	6888	0.2332	99.7512
0.0793	8.0	7872	0.2341	96.8767
0.0708	9.0	8856	0.2409	95.2736
0.0632	10.0	9840	0.2500	100.0829

Framework versions

• Transformers 4.47.0
• Pytorch 2.3.1
• Datasets 3.1.0
• Tokenizers 0.21.0