torgo_xlsr_finetune-M04-2

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.2407
• Wer: 1.2835

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
22.4872	0.88	500	3.2706	1.0
3.361	1.75	1000	2.8365	1.0
2.8532	2.63	1500	2.7444	1.0
2.5391	3.5	2000	2.1105	1.0824
1.6217	4.38	2500	1.7736	1.6424
1.1107	5.25	3000	1.5937	1.4918
0.8277	6.13	3500	1.5655	1.4729
0.6872	7.01	4000	1.6192	1.4671
0.5597	7.88	4500	1.6735	1.4176
0.4942	8.76	5000	1.5915	1.3847
0.4447	9.63	5500	1.8509	1.4506
0.3967	10.51	6000	1.7833	1.3929
0.3596	11.38	6500	2.0147	1.3776
0.3409	12.26	7000	1.8649	1.4
0.3169	13.13	7500	1.8252	1.3541
0.2962	14.01	8000	2.1108	1.3906
0.2934	14.89	8500	1.8004	1.3188
0.2564	15.76	9000	1.8681	1.3659
0.2447	16.64	9500	1.9341	1.3318
0.2248	17.51	10000	2.0251	1.3259
0.2234	18.39	10500	1.9982	1.2988
0.1955	19.26	11000	2.0277	1.3024
0.1882	20.14	11500	2.0001	1.2882
0.2022	21.02	12000	1.9842	1.2988
0.163	21.89	12500	1.9931	1.32
0.1732	22.77	13000	2.0577	1.2659
0.1522	23.64	13500	2.0511	1.2812
0.1367	24.52	14000	2.0308	1.2671
0.1393	25.39	14500	2.2392	1.2788
0.1407	26.27	15000	2.1329	1.2824
0.1244	27.15	15500	2.0721	1.2694
0.116	28.02	16000	2.1656	1.2824
0.125	28.9	16500	2.2338	1.2882
0.1063	29.77	17000	2.2407	1.2835

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1+cu116
• Datasets 1.18.3
• Tokenizers 0.13.2