k2e-20s_asr-scr_w2v2-large-lv60_001

This model is a fine-tuned version of facebook/wav2vec2-large-lv60 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 3.9960
• Per: 0.9491
• Pcc: 0.4581
• Ctc Loss: 2.9538
• Mse Loss: 1.3478

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: 1
• seed: 2222
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2235
• training_steps: 22350
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Per	Pcc	Ctc Loss	Mse Loss
30.007	3.0	2235	4.9713	0.9890	0.4691	3.8731	1.1616
4.5436	6.01	4470	4.8907	0.9890	0.4458	3.7192	1.3424
4.1973	9.01	6705	4.7504	0.9890	0.4791	3.6454	1.3436
3.9659	12.02	8940	4.9631	0.9627	0.3953	3.5684	1.6631
3.7945	15.02	11175	4.6238	0.9627	0.3228	3.5522	1.3937
3.657	18.02	13410	4.8315	0.9627	0.3795	3.4947	1.6564
3.4943	21.03	15645	4.5083	0.9626	0.4295	3.3623	1.4824
3.3082	24.03	17880	4.1212	0.9625	0.4469	3.1651	1.2958
3.1432	27.04	20115	4.1271	0.9586	0.4566	3.0120	1.4192
3.0438	30.04	22350	3.9960	0.9491	0.4581	2.9538	1.3478

Framework versions

• Transformers 4.38.1
• Pytorch 2.0.1
• Datasets 2.16.1
• Tokenizers 0.15.2