whisper-tiny-khmer-aug-v2

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: 0.2740
• Wer: 69.5152

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
0.8546	0.9993	766	0.4256	86.9629
0.3717	2.0	1533	0.3108	82.3739
0.283	2.9993	2299	0.2706	74.4446
0.2343	4.0	3066	0.2663	75.4338
0.2058	4.9993	3832	0.2580	71.1043
0.1805	6.0	4599	0.2582	69.7097
0.1608	6.9993	5365	0.2508	69.9530
0.1437	8.0	6132	0.2586	67.6504
0.1309	8.9993	6898	0.2634	74.7851
0.1206	9.9935	7660	0.2740	69.5152

Framework versions

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