wav2vec2-Y_speed_freq2

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: 113.5747
• Cer: 139.4267

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
137.9274	0.1289	200	113.5748	139.3973
141.0753	0.2579	400	113.5747	139.0977
135.8834	0.3868	600	113.5748	139.2211
144.9369	0.5158	800	113.5747	139.7086
138.9605	0.6447	1000	113.5746	139.3503
133.949	0.7737	1200	113.5748	139.3386
140.5077	0.9026	1400	113.5747	139.5794
136.6605	1.0316	1600	113.5746	139.2211
139.7982	1.1605	1800	113.5748	139.4091
132.729	1.2895	2000	113.5747	139.5089
143.8527	1.4184	2200	113.5747	139.2857
139.8027	1.5474	2400	113.5747	139.3562
134.4609	1.6763	2600	113.5747	139.3797
134.4437	1.8053	2800	113.5748	139.3562
142.3518	1.9342	3000	113.5747	139.2681
138.2023	2.0632	3200	113.5746	139.3268
136.824	2.1921	3400	113.5747	139.3738
136.6418	2.3211	3600	113.5746	139.5031
135.6343	2.4500	3800	113.5748	139.5089
140.8444	2.5790	4000	113.5746	139.1917
140.4969	2.7079	4200	113.5748	139.0566
135.8527	2.8369	4400	113.5747	139.4502
144.0101	2.9658	4600	113.5747	139.4267

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.0+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3