README.md

metadata

library_name: transformers
language:
  - bem
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
  - generated_from_trainer
datasets:
  - BIG_C/BEMBA
metrics:
  - wer
model-index:
  - name: facebook/wav2vec2-xls-r-300m
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: BIG_C
          type: BIG_C/BEMBA
        metrics:
          - name: Wer
            type: wer
            value: 0.49901363753323613

facebook/wav2vec2-xls-r-300m

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the BIG_C dataset. It achieves the following results on the evaluation set:

• Loss: 0.4942
• Wer: 0.4990
• Cer: 0.1073

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: 0.0007
• train_batch_size: 16
• eval_batch_size: 4
• 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: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
1.0121	0.9998	2571	0.5663	0.5221	0.1352
0.605	2.0	5143	0.5692	0.5190	0.1370
0.5815	2.9998	7714	0.5447	0.5001	0.1304
0.5597	4.0	10286	0.5416	0.4768	0.1230
0.5399	4.9998	12857	0.5405	0.5097	0.1381
0.5252	6.0	15429	0.5133	0.4555	0.1184
0.508	6.9998	18000	0.5061	0.4587	0.1192
0.4906	8.0	20572	0.5022	0.4432	0.1145
0.4811	8.9998	23143	0.5004	0.4577	0.1197
0.4628	10.0	25715	0.4940	0.4229	0.1115
0.4499	10.9998	28286	0.4915	0.4067	0.1091
0.4363	12.0	30858	0.4915	0.4205	0.1141
0.4207	12.9998	33429	0.4974	0.4089	0.1096
0.41	14.0	36001	0.5056	0.4079	0.1111
0.393	14.9998	38572	0.4921	0.4132	0.1099
0.3811	16.0	41144	0.4877	0.4127	0.1107
0.3646	16.9998	43715	0.5044	0.3893	0.1048
0.3534	18.0	46287	0.5062	0.3975	0.1062
0.3378	18.9998	48858	0.5130	0.3873	0.1055
0.3206	20.0	51430	0.5048	0.3857	0.1035
0.3063	20.9998	54001	0.4980	0.3958	0.1089
0.2875	22.0	56573	0.5180	0.3856	0.1024
0.2691	22.9998	59144	0.5409	0.3798	0.1034
0.2531	24.0	61716	0.5352	0.4047	0.1087
0.2347	24.9998	64287	0.5459	0.3972	0.1079
0.2156	26.0	66859	0.6152	0.3842	0.1029
0.1996	26.9998	69430	0.6398	0.3802	0.1024
0.1816	28.0	72002	0.6607	0.3849	0.1037
0.1646	28.9998	74573	0.6634	0.3908	0.1055
0.1509	30.0	77145	0.7206	0.4000	0.1071
0.1368	30.9998	79716	0.8071	0.3844	0.1028
0.125	32.0	82288	0.7391	0.4300	0.1171
0.1136	32.9998	84859	0.8141	0.3874	0.1039

Framework versions

• Transformers 4.45.2
• Pytorch 2.1.0+cu118
• Datasets 3.0.1
• Tokenizers 0.20.1