CoolFace
Modelpublic

DrishtiSharma/wav2vec2-large-xls-r-300m-maltese

sourceHugging Faceapache-2.0updated 5y agoView on Hugging Face
0likes86downloads
Model Card

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

wav2vec2-large-xls-r-300m-maltese

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the MOZILLA-FOUNDATION/COMMONVOICE8_0 - MT dataset. It achieves the following results on the evaluation set:

  • —Loss: 0.2994
  • —Wer: 0.2781

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: 7e-05
  • —trainbatchsize: 32
  • —evalbatchsize: 16
  • —seed: 42
  • —optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • —lrschedulertype: linear
  • —lrschedulerwarmup_steps: 1800
  • —num_epochs: 100.0
  • —mixedprecisiontraining: Native AMP

Training results

Training LossEpochStepValidation LossWer
3.01749.0110003.05521.0
1.044618.0220000.67080.7577
0.799527.0330000.42020.4770
0.697836.0440000.30540.3494
0.618945.0550000.28780.3154
0.566754.0560000.31140.3286
0.517363.0670000.30850.3021
0.468272.0780000.30580.2969
0.45181.0890000.31460.2907
0.421390.09100000.30300.2881
0.400599.1110000.30010.2789

Framework versions

  • —Transformers 4.17.0.dev0
  • —Pytorch 1.10.2+cu102
  • —Datasets 1.18.2.dev0
  • —Tokenizers 0.11.0

Evaluation Script

!python eval.py \ --modelid DrishtiSharma/wav2vec2-large-xls-r-300m-maltese \ --dataset mozilla-foundation/commonvoice80 --config mt --split test --log_outputs