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

wav2vec2-xls-r-Wolof-10-hours-ALFFA-dataset

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

  • Loss: 0.2986
  • Wer: 0.1580
  • Cer: 0.0428

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: 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: 0.1
  • num_epochs: 50

Training results

Training Loss Epoch Step Validation Loss Wer Cer
5.4583 1.5936 400 3.0178 1.0 1.0
2.6306 3.1873 800 0.7416 0.7123 0.1944
0.7026 4.7809 1200 0.3958 0.4148 0.1146
0.5091 6.3745 1600 0.3209 0.3380 0.0903
0.4265 7.9681 2000 0.2827 0.2920 0.0775
0.359 9.5618 2400 0.2793 0.2779 0.0756
0.309 11.1554 2800 0.2650 0.2605 0.0692
0.2658 12.7490 3200 0.2715 0.2609 0.0699
0.2315 14.3426 3600 0.2830 0.2498 0.0672
0.2069 15.9363 4000 0.2651 0.2370 0.0623
0.179 17.5299 4400 0.3001 0.2328 0.0641
0.1561 19.1235 4800 0.3151 0.2238 0.0609
0.1438 20.7171 5200 0.2871 0.2160 0.0585
0.1264 22.3108 5600 0.2987 0.2067 0.0564
0.1197 23.9044 6000 0.2798 0.1952 0.0532
0.1093 25.4980 6400 0.2960 0.1969 0.0539
0.1037 27.0916 6800 0.3314 0.2068 0.0563
0.0947 28.6853 7200 0.2852 0.1892 0.0514
0.0917 30.2789 7600 0.3085 0.1935 0.0524
0.0842 31.8725 8000 0.3068 0.1897 0.0516
0.079 33.4661 8400 0.3181 0.1880 0.0506
0.0729 35.0598 8800 0.3027 0.1854 0.0488
0.0719 36.6534 9200 0.3023 0.1778 0.0477
0.0666 38.2470 9600 0.3163 0.1822 0.0487
0.062 39.8406 10000 0.3158 0.1746 0.0476
0.0609 41.4343 10400 0.3172 0.1731 0.0461
0.0561 43.0279 10800 0.3022 0.1639 0.0440
0.0519 44.6215 11200 0.3067 0.1667 0.0446
0.0508 46.2151 11600 0.3068 0.1628 0.0436
0.0491 47.8088 12000 0.3001 0.1599 0.0427
0.0458 49.4024 12400 0.2986 0.1580 0.0428

Framework versions

  • Transformers 4.44.1
  • Pytorch 2.1.0+cu118
  • Datasets 2.17.0
  • Tokenizers 0.19.1
Downloads last month
0
Safetensors
Model size
315M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for asr-africa/wav2vec2-xls-r-Wolof-10-hours-ALFFA-dataset

Finetuned
(524)
this model