You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this model content.

Visualize in Weights & Biases

w2v2-bert-Wolof-10-hours-ALFFA-dataset

This model is a fine-tuned version of facebook/w2v-bert-2.0 on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5709
  • Wer: 0.1743
  • Cer: 0.0523

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: 0.0003
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 4
  • 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: 0.1
  • num_epochs: 50

Training results

Training Loss Epoch Step Validation Loss Wer Cer
1.2343 1.5952 400 0.3992 0.3720 0.0987
0.5062 3.1904 800 0.3984 0.3519 0.0964
0.5063 4.7856 1200 0.5177 0.4211 0.1236
0.4841 6.3809 1600 0.5139 0.3837 0.1151
0.4184 7.9761 2000 0.4656 0.3576 0.1044
0.3573 9.5713 2400 0.4096 0.3080 0.0892
0.3076 11.1665 2800 0.3907 0.2882 0.0836
0.2697 12.7617 3200 0.4390 0.3265 0.0963
0.2364 14.3569 3600 0.3975 0.2941 0.0882
0.2081 15.9521 4000 0.3985 0.2907 0.0863
0.1724 17.5474 4400 0.3945 0.2676 0.0806
0.1502 19.1426 4800 0.4333 0.2634 0.0824
0.1239 20.7378 5200 0.3864 0.2283 0.0702
0.0988 22.3330 5600 0.3749 0.2349 0.0709
0.0832 23.9282 6000 0.3701 0.2270 0.0692
0.0662 25.5234 6400 0.3671 0.2215 0.0665
0.0553 27.1186 6800 0.4373 0.2151 0.0651
0.0436 28.7139 7200 0.4348 0.2153 0.0643
0.0344 30.3091 7600 0.4954 0.2245 0.0700
0.0271 31.9043 8000 0.3983 0.2007 0.0604
0.0196 33.4995 8400 0.4608 0.2199 0.0685
0.0182 35.0947 8800 0.4392 0.1948 0.0590
0.0115 36.6899 9200 0.4944 0.2078 0.0639
0.0104 38.2851 9600 0.4397 0.1910 0.0580
0.0065 39.8804 10000 0.4826 0.1827 0.0549
0.0061 41.4756 10400 0.4912 0.1836 0.0538
0.0045 43.0708 10800 0.4695 0.1859 0.0555
0.0026 44.6660 11200 0.5421 0.1834 0.0556
0.0018 46.2612 11600 0.5372 0.1799 0.0536
0.0008 47.8564 12000 0.5594 0.1768 0.0531
0.0005 49.4516 12400 0.5709 0.1743 0.0523

Framework versions

  • Transformers 4.44.2
  • Pytorch 2.1.0+cu118
  • Datasets 2.17.0
  • Tokenizers 0.19.1
Downloads last month
0
Safetensors
Model size
606M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.

Model tree for asr-africa/w2v2-bert-Wolof-10-hours-ALFFA-dataset

Finetuned
(253)
this model