whisper-base-khmer-aug-v6

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: 0.2379
• Wer: 62.5101

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
0.5956	0.9994	837	0.2717	78.0931
0.2404	2.0	1675	0.2206	79.6660
0.184	2.9994	2512	0.2061	68.4287
0.1511	4.0	3350	0.2001	66.4505
0.1288	4.9994	4187	0.2038	66.2883
0.1108	6.0	5025	0.2032	64.6506
0.0968	6.9994	5862	0.2098	64.0182
0.0842	8.0	6700	0.2180	63.5966
0.0739	8.9994	7537	0.2303	63.9857
0.065	9.9940	8370	0.2379	62.5101

Framework versions

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