finetune_wav2vec2_960h_six_second

This model is a fine-tuned version of facebook/wav2vec2-base-960h on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8664
• Wer: 34.7919
• Cer: 18.1492

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: 32
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2000
• training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.9855	18.5185	1000	0.8664	34.7919	18.1492
0.5055	37.0370	2000	0.9980	34.5251	18.1828
0.3066	55.5556	3000	1.0063	33.3511	17.2474
0.2186	74.0741	4000	1.1086	32.3372	16.9617
0.1628	92.5926	5000	1.1707	31.4835	16.5416
0.1362	111.1111	6000	1.1494	31.2700	16.4351
0.1069	129.6296	7000	1.2482	31.8837	16.4295
0.1004	148.1481	8000	1.3189	31.5635	16.9393
0.0851	166.6667	9000	1.3079	30.8965	16.3343
0.0794	185.1852	10000	1.3297	30.8698	16.1214

Framework versions

• Transformers 4.40.2
• Pytorch 1.12.1+cu116
• Datasets 2.19.1
• Tokenizers 0.19.1