<!-- 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. -->
dgx1_w2v2_base_distillation_att_loss_mozilla_epochs_100_batch_16_concat_data
This model is a fine-tuned version of on the None dataset. It achieves the following results on the evaluation set:
- Loss: 31.8691
- Wer: 0.3078
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 128
- total_train_batch_size: 2048
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.2
- num_epochs: 100
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
78.8961 | 7.35 | 500 | 26.8983 | 0.3072 |
61.0428 | 14.7 | 1000 | 28.0016 | 0.3077 |
57.2458 | 22.06 | 1500 | 29.1592 | 0.3075 |
54.0333 | 29.41 | 2000 | 29.2930 | 0.3088 |
52.0263 | 36.76 | 2500 | 30.8205 | 0.3079 |
50.1754 | 44.12 | 3000 | 31.8691 | 0.3078 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.12.1
- Datasets 2.8.0
- Tokenizers 0.13.2