Configuration Parsing
Warning:
In adapter_config.json: "peft.task_type" must be a string
whisper-large-v2-lug-eng-extended
This model is a fine-tuned version of openai/whisper-large-v2 on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 0.1374
- Wer Eng: 0.028
- Wer Lug: 0.15
- Wer Mean: 0.089
- Cer Eng: 0.013
- Cer Lug: 0.04
- Cer Mean: 0.027
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: 0.001
- train_batch_size: 16
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 10
- training_steps: 5000
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer Eng | Wer Lug | Wer Mean | Cer Eng | Cer Lug | Cer Mean |
---|---|---|---|---|---|---|---|---|---|
0.3746 | 0.02 | 100 | 0.1615 | 0.029 | 0.197 | 0.113 | 0.014 | 0.053 | 0.034 |
0.4082 | 0.04 | 200 | 0.1703 | 0.044 | 0.235 | 0.14 | 0.022 | 0.062 | 0.042 |
0.4169 | 0.06 | 300 | 0.1744 | 0.036 | 0.197 | 0.116 | 0.017 | 0.049 | 0.033 |
0.3956 | 0.08 | 400 | 0.1714 | 0.038 | 0.205 | 0.121 | 0.019 | 0.049 | 0.034 |
0.4083 | 0.1 | 500 | 0.1730 | 0.032 | 0.215 | 0.124 | 0.016 | 0.052 | 0.034 |
0.3946 | 0.12 | 600 | 0.1706 | 0.037 | 0.208 | 0.122 | 0.014 | 0.049 | 0.032 |
0.4188 | 0.14 | 700 | 0.1670 | 0.035 | 0.215 | 0.125 | 0.014 | 0.051 | 0.033 |
0.4372 | 0.16 | 800 | 0.1633 | 0.036 | 0.224 | 0.13 | 0.018 | 0.051 | 0.034 |
0.4245 | 0.18 | 900 | 0.1705 | 0.037 | 0.228 | 0.133 | 0.019 | 0.057 | 0.038 |
0.4222 | 0.2 | 1000 | 0.1642 | 0.032 | 0.206 | 0.119 | 0.015 | 0.05 | 0.033 |
0.4152 | 0.22 | 1100 | 0.1602 | 0.037 | 0.197 | 0.117 | 0.017 | 0.047 | 0.032 |
0.4091 | 0.24 | 1200 | 0.1653 | 0.035 | 0.191 | 0.113 | 0.015 | 0.046 | 0.03 |
0.4183 | 0.26 | 1300 | 0.1607 | 0.037 | 0.188 | 0.113 | 0.019 | 0.045 | 0.032 |
0.3596 | 1.0195 | 1400 | 0.1578 | 0.035 | 0.179 | 0.107 | 0.017 | 0.049 | 0.033 |
0.3566 | 1.0395 | 1500 | 0.1568 | 0.032 | 0.194 | 0.113 | 0.015 | 0.048 | 0.031 |
0.3901 | 1.0595 | 1600 | 0.1579 | 0.041 | 0.187 | 0.114 | 0.018 | 0.043 | 0.031 |
0.3511 | 1.0795 | 1700 | 0.1768 | 0.04 | 0.19 | 0.115 | 0.019 | 0.051 | 0.035 |
0.3997 | 1.0995 | 1800 | 0.1580 | 0.036 | 0.173 | 0.104 | 0.016 | 0.043 | 0.03 |
0.3626 | 1.1195 | 1900 | 0.1583 | 0.035 | 0.169 | 0.102 | 0.015 | 0.043 | 0.029 |
0.372 | 1.1395 | 2000 | 0.1535 | 0.034 | 0.179 | 0.106 | 0.017 | 0.047 | 0.032 |
0.3578 | 1.1595 | 2100 | 0.1528 | 0.031 | 0.197 | 0.114 | 0.016 | 0.05 | 0.033 |
0.3697 | 1.1795 | 2200 | 0.1533 | 0.038 | 0.198 | 0.118 | 0.016 | 0.051 | 0.034 |
0.3651 | 1.1995 | 2300 | 0.1515 | 0.031 | 0.187 | 0.109 | 0.014 | 0.059 | 0.037 |
0.366 | 1.2195 | 2400 | 0.1466 | 0.032 | 0.164 | 0.098 | 0.014 | 0.046 | 0.03 |
0.4013 | 1.2395 | 2500 | 0.1490 | 0.036 | 0.169 | 0.102 | 0.016 | 0.045 | 0.03 |
0.337 | 1.2595 | 2600 | 0.1505 | 0.03 | 0.187 | 0.109 | 0.015 | 0.051 | 0.033 |
0.2846 | 2.019 | 2700 | 0.1533 | 0.034 | 0.177 | 0.105 | 0.015 | 0.044 | 0.03 |
0.2825 | 2.039 | 2800 | 0.1475 | 0.03 | 0.168 | 0.099 | 0.013 | 0.041 | 0.027 |
0.3233 | 2.059 | 2900 | 0.1444 | 0.032 | 0.168 | 0.1 | 0.015 | 0.041 | 0.028 |
0.3075 | 2.079 | 3000 | 0.1416 | 0.029 | 0.171 | 0.1 | 0.012 | 0.042 | 0.027 |
0.2988 | 2.099 | 3100 | 0.1421 | 0.028 | 0.176 | 0.102 | 0.012 | 0.046 | 0.029 |
0.2782 | 2.1190 | 3200 | 0.1414 | 0.026 | 0.16 | 0.093 | 0.011 | 0.044 | 0.027 |
0.3123 | 2.1390 | 3300 | 0.1494 | 0.025 | 0.172 | 0.098 | 0.01 | 0.044 | 0.027 |
0.2839 | 2.159 | 3400 | 0.1443 | 0.03 | 0.155 | 0.093 | 0.015 | 0.04 | 0.027 |
0.2812 | 2.179 | 3500 | 0.1449 | 0.025 | 0.168 | 0.096 | 0.013 | 0.045 | 0.029 |
0.3022 | 2.199 | 3600 | 0.1453 | 0.03 | 0.143 | 0.087 | 0.015 | 0.036 | 0.025 |
0.2753 | 2.219 | 3700 | 0.1428 | 0.025 | 0.164 | 0.094 | 0.01 | 0.042 | 0.026 |
0.3011 | 2.239 | 3800 | 0.1398 | 0.034 | 0.15 | 0.092 | 0.015 | 0.037 | 0.026 |
0.2895 | 2.259 | 3900 | 0.1405 | 0.028 | 0.162 | 0.095 | 0.012 | 0.044 | 0.028 |
0.2503 | 3.0185 | 4000 | 0.1400 | 0.029 | 0.161 | 0.095 | 0.012 | 0.043 | 0.028 |
0.236 | 3.0385 | 4100 | 0.1412 | 0.03 | 0.161 | 0.096 | 0.013 | 0.042 | 0.028 |
0.2366 | 3.0585 | 4200 | 0.1417 | 0.03 | 0.16 | 0.095 | 0.013 | 0.042 | 0.028 |
0.2393 | 3.0785 | 4300 | 0.1396 | 0.028 | 0.147 | 0.088 | 0.013 | 0.037 | 0.025 |
0.2348 | 3.0985 | 4400 | 0.1419 | 0.028 | 0.154 | 0.091 | 0.012 | 0.04 | 0.026 |
0.2221 | 3.1185 | 4500 | 0.1364 | 0.031 | 0.146 | 0.089 | 0.013 | 0.039 | 0.026 |
0.2375 | 3.1385 | 4600 | 0.1393 | 0.028 | 0.149 | 0.088 | 0.014 | 0.038 | 0.026 |
0.2383 | 3.1585 | 4700 | 0.1385 | 0.021 | 0.15 | 0.085 | 0.01 | 0.04 | 0.025 |
0.2269 | 3.1785 | 4800 | 0.1376 | 0.03 | 0.143 | 0.087 | 0.014 | 0.038 | 0.026 |
0.2351 | 3.1985 | 4900 | 0.1372 | 0.028 | 0.157 | 0.092 | 0.013 | 0.041 | 0.027 |
0.2315 | 3.2185 | 5000 | 0.1374 | 0.028 | 0.15 | 0.089 | 0.013 | 0.04 | 0.027 |
Framework versions
- PEFT 0.12.1.dev0
- Transformers 4.45.0.dev0
- Pytorch 2.2.0
- Datasets 2.21.0
- Tokenizers 0.19.1
- Downloads last month
- 7
Model tree for jq/whisper-large-v2-lug-eng-extended
Base model
openai/whisper-large-v2