ASR-cv-corpus-ug-15

This model is a fine-tuned version of piyazon/ASR-cv-corpus-ug-14 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0090
• Wer: 0.0069

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: 3e-05
• train_batch_size: 32
• 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: 300
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0087	0.2338	500	0.0117	0.0126
0.0103	0.4675	1000	0.0108	0.0122
0.0091	0.7013	1500	0.0093	0.0108
0.0099	0.9350	2000	0.0113	0.0150
0.0071	1.1688	2500	0.0099	0.0101
0.0054	1.4025	3000	0.0108	0.0112
0.0057	1.6363	3500	0.0096	0.0112
0.006	1.8700	4000	0.0088	0.0104
0.0046	2.1038	4500	0.0092	0.0110
0.0049	2.3375	5000	0.0095	0.0106
0.0044	2.5713	5500	0.0093	0.0106
0.0043	2.8050	6000	0.0085	0.0098
0.0036	3.0388	6500	0.0088	0.0094
0.0029	3.2726	7000	0.0089	0.0097
0.003	3.5063	7500	0.0085	0.0093
0.0032	3.7401	8000	0.0090	0.0093
0.0029	3.9738	8500	0.0084	0.0090
0.0019	4.2076	9000	0.0093	0.0089
0.0022	4.4413	9500	0.0083	0.0097
0.0022	4.6751	10000	0.0086	0.0092
0.0021	4.9088	10500	0.0085	0.0087
0.002	5.1426	11000	0.0089	0.0090
0.0011	5.3763	11500	0.0079	0.0081
0.0014	5.6101	12000	0.0076	0.0085
0.0014	5.8439	12500	0.0090	0.0090
0.0013	6.0776	13000	0.0082	0.0080
0.0009	6.3114	13500	0.0086	0.0083
0.0009	6.5451	14000	0.0088	0.0084
0.0009	6.7789	14500	0.0079	0.0071
0.0007	7.0126	15000	0.0083	0.0074
0.0006	7.2464	15500	0.0083	0.0081
0.0005	7.4801	16000	0.0092	0.0083
0.0005	7.7139	16500	0.0093	0.0078
0.0006	7.9476	17000	0.0088	0.0077
0.0003	8.1814	17500	0.0089	0.0071
0.0004	8.4151	18000	0.0089	0.0070
0.0004	8.6489	18500	0.0082	0.0071
0.0002	8.8827	19000	0.0086	0.0071
0.0001	9.1164	19500	0.0089	0.0071
0.0002	9.3502	20000	0.0092	0.0071
0.0001	9.5839	20500	0.0090	0.0071
0.0001	9.8177	21000	0.0090	0.0069

Framework versions

• Transformers 4.56.1
• Pytorch 2.8.0+cu128
• Datasets 4.1.0
• Tokenizers 0.22.0