wav2vec2-xlsr-et-lm-1B
This model was finetuned with mozilla_foundation/common_voice_8_0 et with train+other+validation splits. It achieves the following results on the test set: (Loss reported with last eval step at step 2000/2040 during training)
- Loss: 0.2150
- Wer: 0.2012
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: 0.00005
- train_batch_size: 32
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 1
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
Framework versions
- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.3
- Tokenizers 0.11.0
- Downloads last month
- 6
Inference Providers
NEW
This model is not currently available via any of the supported third-party Inference Providers, and
the model is not deployed on the HF Inference API.
Dataset used to train RASMUS/wav2vec2-xlsr-1b-et
Evaluation results
- Test WER on Common Voice 8self-reported20.120
- Test CER on Common Voice 8self-reported3.820
- Test WER on Robust Speech Event - Dev Dataself-reported40.770
- Test CER on Robust Speech Event - Dev Dataself-reported12.320
- Test WER on Robust Speech Event - Test Dataself-reported41.970