metadata

base_model: facebook/mms-1b-all
datasets:
  - common_voice_17_0
library_name: transformers
license: cc-by-nc-4.0
metrics:
  - wer
  - bleu
tags:
  - generated_from_trainer
model-index:
  - name: wav2vec2-mms-1b-CV17.0-training_set_variations
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: common_voice_17_0
          type: common_voice_17_0
          config: ta
          split: validation
          args: ta
        metrics:
          - type: wer
            value: 0.35028609973948643
            name: Wer
          - type: bleu
            value: 0.4335646189418293
            name: Bleu

wav2vec2-mms-1b-CV17.0-training_set_variations

This model is a fine-tuned version of facebook/mms-1b-all on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

Loss: 0.1970
Wer: 0.3503
Cer: 0.0557
Bleu: 0.4336

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: 0.001
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.15
training_steps: 2000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer	Bleu
7.329	0.0977	100	0.3217	0.4338	0.0716	0.3272
0.2154	0.1953	200	0.2314	0.3794	0.0634	0.3999
0.189	0.2930	300	0.2188	0.3656	0.0592	0.4166
0.1975	0.3906	400	0.2212	0.3763	0.0608	0.4032
0.1813	0.4883	500	0.2117	0.3634	0.0585	0.4171
0.1791	0.5859	600	0.2074	0.3590	0.0578	0.4220
0.187	0.6836	700	0.2087	0.3607	0.0582	0.4188
0.1789	0.7812	800	0.2064	0.3542	0.0568	0.4327
0.1704	0.8789	900	0.2076	0.3661	0.0587	0.4095
0.1813	0.9766	1000	0.2044	0.3589	0.0574	0.4200
0.1633	1.0742	1100	0.2029	0.3582	0.0575	0.4220
0.1699	1.1719	1200	0.2034	0.3537	0.0566	0.4335
0.1822	1.2695	1300	0.2037	0.3589	0.0578	0.4227
0.1654	1.3672	1400	0.2028	0.3549	0.0568	0.4288
0.1696	1.4648	1500	0.2011	0.3579	0.0567	0.4199
0.1622	1.5625	1600	0.1999	0.3568	0.0570	0.4228
0.1742	1.6602	1700	0.1983	0.3490	0.0559	0.4365
0.1581	1.7578	1800	0.1973	0.3511	0.0558	0.4329
0.1616	1.8555	1900	0.1970	0.3482	0.0556	0.4381
0.1607	1.9531	2000	0.1970	0.3503	0.0557	0.4336

Framework versions

Transformers 4.44.2
Pytorch 2.4.1+cu121
Datasets 3.0.0
Tokenizers 0.19.1