This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the generator dataset. It achieves the following results on the evaluation set:

Loss: 0.3020
Wer: 0.2061

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: 4
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 16
optimizer: Use OptimizerNames.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: 1000
training_steps: 1600

Training results

Training Loss	Step	Validation Loss	Wer
9.7238	400	4.3969	1.0000
4.7833	800	3.4483	1.0000
3.407	1200	3.3647	1.0000
2.3162	1600	1.6377	0.9461
1.7377	2000	1.2870	0.8850
1.4824	2400	1.0638	0.6756
1.3022	2800	0.9677	0.6144
1.2161	3200	0.9012	0.5805
1.1507	3600	0.8627	0.5556
1.1122	4000	0.8179	0.5373
1.0667	4400	0.8080	0.5243
1.0605	4800	0.7947	0.5131
1.0229	5200	0.7955	0.5057
1.0302	5600	0.7778	0.5015
1.0061	6000	0.7866	0.4910
0.9727	6400	0.7485	0.4661
0.9473	6800	0.7261	0.4480
0.9033	7200	0.6793	0.4289
0.8801	7600	0.6727	0.4187
0.841	8000	0.6516	0.4098
0.8444	8400	0.6555	0.4033
0.7837	8800	0.6274	0.3709
0.7521	9200	0.6054	0.3651
0.7347	9600	0.5874	0.3617
0.7016	10000	0.5702	0.3542
0.7089	10400	0.5749	0.3491
0.6764	10800	0.5636	0.3443
0.6805	11200	0.5645	0.3392
0.6748	11600	0.5781	0.3363
0.6755	12000	0.5584	0.3366
0.6588	12400	0.5599	0.3315
0.6725	12800	0.5585	0.3280
0.6542	13200	0.5621	0.3264
0.6718	13600	0.5584	0.3267
0.6193	14000	0.5056	0.3152
0.5927	14400	0.5112	0.2921
0.5711	14800	0.5021	0.2926
0.5655	15200	0.5047	0.2842
0.5578	15600	0.4958	0.2830
0.4854	16000	0.3020	0.2461
0.4780	16400	0.2932	0.2321
0.4802	16800	0.2976	0.2373
0.4732	17200	0.2906	0.2290
0.4701	17600	0.2882	0.2225
0.4642	18000	0.2834	0.2163
0.4514	18400	0.2784	0.2061
0.4732	18800	0.2906	0.2290
0.4712	19200	0.2882	0.2225
0.4651	19600	0.2834	0.2163
0.4704	19200	0.2882	0.2225
0.4679	19600	0.2834	0.2193

Framework versions

Transformers 4.46.3
Pytorch 2.5.1+cu121
Datasets 4.4.1
Tokenizers 0.20.3

pdabo1607/English_Wav2Vec_Finetune

English_Wav2Vec_Finetune

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

pdabo1607/
English_Wav2Vec_Finetune