fs-w-xavier-tiny

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: 0.4490
• Wer: 106.9801
• Cer: 82.1968

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: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
4.2145	4.5872	500	4.2399	95.4891	77.5678
1.4437	9.1743	1000	1.6923	107.8348	86.0529
0.5423	13.7615	1500	0.8639	109.7341	86.0014
0.3798	18.3486	2000	0.6183	106.5052	85.3830
0.3301	22.9358	2500	0.5421	114.9098	92.5713
0.2883	27.5229	3000	0.4921	100.9022	76.8722
0.2381	32.1101	3500	0.4725	108.4520	85.7953
0.2115	36.6972	4000	0.4611	103.2289	79.1996
0.1983	41.2844	4500	0.4507	104.8908	81.3037
0.1668	45.8716	5000	0.4490	106.9801	82.1968

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0