metadata

license: cc-by-nc-4.0
base_model: utter-project/mHuBERT-147
tags:
  - generated_from_trainer
datasets:
  - common_voice_15_0
metrics:
  - wer
model-index:
  - name: mHuBERT-147-br
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: common_voice_15_0
          type: common_voice_15_0
          config: br
          split: None
          args: br
        metrics:
          - name: Wer
            type: wer
            value: 54.40414507772021

mHuBERT-147-br

This model is a fine-tuned version of utter-project/mHuBERT-147 on the common_voice_15_0 dataset. It achieves the following results on the evaluation set:

Loss: 0.7748
Wer: 54.4041
Cer: 18.4091

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: 3.5e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 16
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 40
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
6.9168	2.18	1000	3.4435	100.0	99.8848
2.5187	4.36	2000	1.5458	84.7983	31.7071
1.2569	6.54	3000	1.0204	75.0740	26.1506
0.9322	8.71	4000	0.8765	69.9852	24.0654
0.785	10.89	5000	0.8191	66.0252	22.4968
0.6997	13.07	6000	0.8166	64.1007	21.8478
0.6318	15.25	7000	0.7961	61.4730	20.9685
0.5827	17.43	8000	0.7853	59.9926	20.2523
0.5573	19.61	9000	0.7536	59.6873	20.0737
0.5173	21.79	10000	0.7525	58.3364	19.6014
0.4874	23.97	11000	0.7694	57.4759	19.4766
0.4643	26.14	12000	0.7800	56.1158	19.0984
0.4511	28.32	13000	0.7640	55.6255	18.7892
0.4268	30.5	14000	0.7495	55.4404	18.6548
0.423	32.68	15000	0.7641	55.0703	18.5281
0.4166	34.86	16000	0.7730	54.8020	18.5377
0.3968	37.04	17000	0.7658	54.4597	18.3995
0.3958	39.22	18000	0.7748	54.4041	18.4091

Framework versions

Transformers 4.39.1
Pytorch 2.0.1+cu117
Datasets 2.18.0
Tokenizers 0.15.2