whisper-tiny-minds14-en

This model is a fine-tuned version of openai/whisper-tiny on the PolyAI/minds14 dataset. It achieves the following results on the evaluation set:

• Loss: 0.5435
• Wer Ortho: 0.3455
• Wer: 0.3318

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: 1e-05
• train_batch_size: 8
• eval_batch_size: 8
• 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: 50
• training_steps: 500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
3.1062	0.4386	25	2.2016	0.5077	0.3955
1.3065	0.8772	50	0.5662	0.4183	0.3878
0.4383	1.3158	75	0.4933	0.3775	0.3613
0.4020	1.7544	100	0.4771	0.3578	0.3453
0.3583	2.1930	125	0.4733	0.3689	0.3571
0.2264	2.6316	150	0.4765	0.3689	0.3583
0.2011	3.0702	175	0.4696	0.3350	0.3235
0.1494	3.5088	200	0.4826	0.3387	0.3241
0.1448	3.9474	225	0.4852	0.3535	0.3394
0.0698	4.3860	250	0.4920	0.3251	0.3146
0.0871	4.8246	275	0.5013	0.3257	0.3140
0.0560	5.2632	300	0.5130	0.3331	0.3217
0.0414	5.7018	325	0.5216	0.3430	0.3323
0.0347	6.1404	350	0.5242	0.3362	0.3247
0.0205	6.5789	375	0.5344	0.3313	0.3205
0.0259	7.0175	400	0.5328	0.3436	0.3335
0.0122	7.4561	425	0.5374	0.3467	0.3365
0.0213	7.8947	450	0.5417	0.3455	0.3329
0.0102	8.3333	475	0.5428	0.3424	0.3282
0.0111	8.7719	500	0.5435	0.3455	0.3318

Framework versions

• Transformers 5.3.0
• Pytorch 2.10.0+cu128
• Datasets 4.8.3
• Tokenizers 0.22.2