whisper-tiny-khmer-aug-v6

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

• Loss: 0.2805
• Wer: 101.6307

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7315	1.0	984	0.3818	100.9121
0.2869	2.0	1968	0.2958	100.8292
0.218	3.0	2952	0.2682	121.5589
0.1864	4.0	3936	0.2559	102.4323
0.1618	5.0	4920	0.2580	100.6081
0.145	6.0	5904	0.2583	109.2869
0.1306	7.0	6888	0.2693	113.9856
0.1192	8.0	7872	0.2624	101.8795
0.1095	9.0	8856	0.2723	97.9823
0.1001	10.0	9840	0.2805	101.6307

Framework versions

• Transformers 4.44.0
• Pytorch 2.3.1
• Datasets 2.21.0
• Tokenizers 0.19.1