metadata
license: apache-2.0
tags:
- automatic-speech-recognition
- experiments/data/uwb_atcc/train
- generated_from_trainer
metrics:
- wer
model-index:
- name: 0.0ld_0.0ad_0.0attd_0.05fpd_0.075mtp_12mtl_0.0mfp_12mfl_1acc
results: []
0.0ld_0.0ad_0.0attd_0.05fpd_0.075mtp_12mtl_0.0mfp_12mfl_1acc
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the EXPERIMENTS/DATA/UWB_ATCC/TRAIN - NA dataset. It achieves the following results on the evaluation set:
- Loss: 0.8470
- Wer: 0.1898
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: 0.0001
- train_batch_size: 24
- eval_batch_size: 12
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- training_steps: 10000
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
No log | 1.06 | 500 | 3.1697 | 1.0 |
3.1489 | 2.12 | 1000 | 1.4184 | 0.5678 |
3.1489 | 3.18 | 1500 | 0.8498 | 0.3366 |
0.8499 | 4.25 | 2000 | 0.8089 | 0.2755 |
0.8499 | 5.31 | 2500 | 0.7339 | 0.2963 |
0.5901 | 6.37 | 3000 | 0.6376 | 0.2402 |
0.5901 | 7.43 | 3500 | 0.6890 | 0.2336 |
0.4724 | 8.49 | 4000 | 0.6844 | 0.2240 |
0.4724 | 9.55 | 4500 | 0.6900 | 0.2222 |
0.3981 | 10.62 | 5000 | 0.7051 | 0.2123 |
0.3981 | 11.68 | 5500 | 0.6671 | 0.2095 |
0.3436 | 12.74 | 6000 | 0.7425 | 0.2049 |
0.3436 | 13.8 | 6500 | 0.7135 | 0.1994 |
0.2925 | 14.86 | 7000 | 0.7350 | 0.2012 |
0.2925 | 15.92 | 7500 | 0.7855 | 0.1945 |
0.2525 | 16.99 | 8000 | 0.7933 | 0.1946 |
0.2525 | 18.05 | 8500 | 0.8016 | 0.1915 |
0.2285 | 19.11 | 9000 | 0.8284 | 0.1907 |
0.2285 | 20.17 | 9500 | 0.8275 | 0.1902 |
0.2025 | 21.23 | 10000 | 0.8470 | 0.1898 |
Framework versions
- Transformers 4.24.0
- Pytorch 1.13.0+cu117
- Datasets 2.6.1
- Tokenizers 0.13.2