2_1000_1e-5_hp-mehrdad
This model is a fine-tuned version of lnxdx/21_2500_1e-4_hp-mehrdad on the None dataset. It achieves the following results on the evaluation set:
- Loss on ShEMO train set: 0.6809
- Loss on ShEMO dev set: 0.6591
- WER on ShEMO train set: 27.41
- WER on ShEMO dev set: 31.37 (Why not 31.36?)
- WER on Common Voice 13 test set: 19.26
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: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- 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
- training_steps: 1000
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.7624 | 0.62 | 100 | 0.6708 | 0.3236 |
| 0.78 | 1.25 | 200 | 0.6668 | 0.3245 |
| 0.7856 | 1.88 | 300 | 0.6600 | 0.3274 |
| 0.7239 | 2.5 | 400 | 0.6672 | 0.3233 |
| 0.7311 | 3.12 | 500 | 0.6748 | 0.3143 |
| 0.7408 | 3.75 | 600 | 0.6518 | 0.3248 |
| 0.713 | 4.38 | 700 | 0.6587 | 0.3178 |
| 0.7068 | 5.0 | 800 | 0.6600 | 0.3172 |
| 0.6938 | 5.62 | 900 | 0.6598 | 0.3157 |
| 0.6809 | 6.25 | 1000 | 0.6591 | 0.3137 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu118
- Datasets 2.15.0
- Tokenizers 0.15.0