tun_msa_wav2vec4

This model was trained from scratch on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.0258
• Wer: 0.7054
• Cer: 0.2514

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-06
• 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: 700
• num_epochs: 60

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
4.9443	6.7669	900	2.9731	0.9978	0.9143
2.385	13.5338	1800	1.4747	0.8366	0.3285
1.6027	20.3008	2700	1.1810	0.7617	0.2766
1.3459	27.0677	3600	1.0963	0.7317	0.2615
1.2812	33.8346	4500	1.0580	0.7240	0.2563
1.2156	40.6015	5400	1.0426	0.7127	0.2538
1.1119	47.3684	6300	1.0305	0.7069	0.2519
1.0993	54.1353	7200	1.0258	0.7054	0.2514

Framework versions

• Transformers 4.41.2
• Pytorch 2.1.2
• Datasets 2.19.2
• Tokenizers 0.19.1