whisper-tiny-amh-matewosx

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: 0.1374
• Wer: 0.5453
• Cer: 0.3897

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: 32
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.3381	0.4310	500	0.3049	0.9142	0.6918
0.2324	0.8621	1000	0.1985	0.6610	0.4449
0.1947	1.2931	1500	0.1737	0.6254	0.4277
0.1656	1.7241	2000	0.1545	0.5950	0.4151
0.1518	2.1552	2500	0.1481	0.5810	0.4079
0.1410	2.5862	3000	0.1417	0.5718	0.4040
0.1312	3.0172	3500	0.1371	0.5610	0.3964
0.1199	3.4483	4000	0.1338	0.5558	0.3973
0.1219	3.8793	4500	0.1297	0.5512	0.3947
0.1084	4.3103	5000	0.1311	0.5485	0.3924
0.1069	4.7414	5500	0.1295	0.5493	0.3914
0.0913	5.1724	6000	0.1326	0.5476	0.3933
0.0920	5.6034	6500	0.1341	0.5503	0.3950
0.0736	6.0345	7000	0.1374	0.5453	0.3897

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2