wav2vec2-Y_freq_speed_pause2

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.7792
• Cer: 61.9389

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch 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
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
33.1457	0.1289	200	4.9717	100.0
5.2051	0.2579	400	4.6998	100.0
4.9209	0.3868	600	4.7110	100.0
4.8854	0.5158	800	4.6652	100.0
4.779	0.6447	1000	4.6563	100.0
4.7482	0.7737	1200	4.6129	100.0
4.7017	0.9026	1400	4.5656	100.0
4.5268	1.0316	1600	4.5651	97.4148
4.2095	1.1605	1800	4.5623	98.3079
3.6552	1.2895	2000	4.1718	85.0588
3.2349	1.4184	2200	3.7242	79.8120
2.9414	1.5474	2400	3.4438	70.0059
2.7299	1.6763	2600	3.2995	69.0188
2.546	1.8053	2800	3.2953	69.2127
2.3883	1.9342	3000	3.2021	69.5182
2.265	2.0632	3200	3.1229	67.0858
2.1059	2.1921	3400	2.9669	65.1351
2.0255	2.3211	3600	2.8778	63.5781
1.98	2.4500	3800	2.8908	63.9894
1.9373	2.5790	4000	2.8894	63.5605
1.8778	2.7079	4200	2.8307	62.5499
1.825	2.8369	4400	2.7265	61.0400
1.7932	2.9658	4600	2.7792	61.9389

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3