<!-- 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. -->
pure-start-epoch1
This model is a fine-tuned version of yongjian/wav2vec2-large-a on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 21.0050
- Acc: 0.095
- Wer: 1.0
- Correct: 19
- Total: 200
- Strlen: 200
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: 9e-06
- train_batch_size: 2
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 8
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 1.0
Training results
Training Loss | Epoch | Step | Validation Loss | Acc | Wer | Correct | Total | Strlen |
---|---|---|---|---|---|---|---|---|
No log | 0.02 | 5 | 67.2752 | 0.0 | 1.0119 | 0 | 200 | 200 |
131.0548 | 0.04 | 10 | 66.2796 | 0.0 | 1.0257 | 0 | 200 | 200 |
131.0548 | 0.06 | 15 | 65.2071 | 0.005 | 1.0237 | 1 | 200 | 200 |
145.0859 | 0.08 | 20 | 64.0987 | 0.035 | 1.0198 | 7 | 200 | 200 |
145.0859 | 0.11 | 25 | 62.9734 | 0.07 | 1.0119 | 14 | 200 | 200 |
110.0012 | 0.13 | 30 | 61.8288 | 0.09 | 1.0119 | 18 | 200 | 200 |
110.0012 | 0.15 | 35 | 60.6565 | 0.09 | 1.0119 | 18 | 200 | 200 |
122.6164 | 0.17 | 40 | 59.4606 | 0.095 | 1.0119 | 19 | 200 | 200 |
122.6164 | 0.19 | 45 | 58.2224 | 0.095 | 1.0099 | 19 | 200 | 200 |
125.942 | 0.21 | 50 | 56.9514 | 0.095 | 1.0020 | 19 | 200 | 200 |
125.942 | 0.23 | 55 | 55.5923 | 0.095 | 1.0 | 19 | 200 | 200 |
111.2271 | 0.25 | 60 | 54.1423 | 0.095 | 1.0 | 19 | 200 | 200 |
111.2271 | 0.27 | 65 | 52.6174 | 0.095 | 1.0 | 19 | 200 | 200 |
137.2356 | 0.3 | 70 | 51.0340 | 0.095 | 1.0 | 19 | 200 | 200 |
137.2356 | 0.32 | 75 | 49.4034 | 0.095 | 1.0 | 19 | 200 | 200 |
112.2532 | 0.34 | 80 | 47.7291 | 0.095 | 1.0 | 19 | 200 | 200 |
112.2532 | 0.36 | 85 | 46.0281 | 0.095 | 1.0 | 19 | 200 | 200 |
88.3973 | 0.38 | 90 | 44.2361 | 0.095 | 1.0 | 19 | 200 | 200 |
88.3973 | 0.4 | 95 | 42.4925 | 0.095 | 1.0 | 19 | 200 | 200 |
88.7175 | 0.42 | 100 | 40.7673 | 0.095 | 1.0 | 19 | 200 | 200 |
88.7175 | 0.44 | 105 | 39.0848 | 0.095 | 1.0 | 19 | 200 | 200 |
90.857 | 0.46 | 110 | 37.4890 | 0.095 | 1.0 | 19 | 200 | 200 |
90.857 | 0.48 | 115 | 35.8966 | 0.095 | 1.0 | 19 | 200 | 200 |
77.5782 | 0.51 | 120 | 34.2822 | 0.1 | 1.0 | 20 | 200 | 200 |
77.5782 | 0.53 | 125 | 32.7953 | 0.1 | 1.0 | 20 | 200 | 200 |
80.2378 | 0.55 | 130 | 31.4560 | 0.1 | 1.0 | 20 | 200 | 200 |
80.2378 | 0.57 | 135 | 30.1651 | 0.1 | 1.0 | 20 | 200 | 200 |
73.5042 | 0.59 | 140 | 29.0069 | 0.095 | 1.0 | 19 | 200 | 200 |
73.5042 | 0.61 | 145 | 28.0349 | 0.095 | 1.0 | 19 | 200 | 200 |
71.5632 | 0.63 | 150 | 27.1812 | 0.095 | 1.0 | 19 | 200 | 200 |
71.5632 | 0.65 | 155 | 26.4012 | 0.095 | 1.0 | 19 | 200 | 200 |
76.5337 | 0.67 | 160 | 25.6924 | 0.095 | 1.0 | 19 | 200 | 200 |
76.5337 | 0.7 | 165 | 25.0184 | 0.095 | 1.0 | 19 | 200 | 200 |
54.6507 | 0.72 | 170 | 24.4100 | 0.095 | 1.0 | 19 | 200 | 200 |
54.6507 | 0.74 | 175 | 23.8273 | 0.095 | 1.0 | 19 | 200 | 200 |
57.1606 | 0.76 | 180 | 23.2988 | 0.095 | 1.0 | 19 | 200 | 200 |
57.1606 | 0.78 | 185 | 22.8731 | 0.095 | 1.0 | 19 | 200 | 200 |
56.0855 | 0.8 | 190 | 22.5336 | 0.095 | 1.0 | 19 | 200 | 200 |
56.0855 | 0.82 | 195 | 22.2334 | 0.095 | 1.0 | 19 | 200 | 200 |
55.2475 | 0.84 | 200 | 21.9555 | 0.095 | 1.0 | 19 | 200 | 200 |
55.2475 | 0.86 | 205 | 21.7112 | 0.095 | 1.0 | 19 | 200 | 200 |
47.9988 | 0.89 | 210 | 21.5123 | 0.095 | 1.0 | 19 | 200 | 200 |
47.9988 | 0.91 | 215 | 21.3407 | 0.095 | 1.0 | 19 | 200 | 200 |
55.1394 | 0.93 | 220 | 21.1965 | 0.095 | 1.0 | 19 | 200 | 200 |
55.1394 | 0.95 | 225 | 21.1028 | 0.095 | 1.0 | 19 | 200 | 200 |
48.0323 | 0.97 | 230 | 21.0376 | 0.095 | 1.0 | 19 | 200 | 200 |
48.0323 | 0.99 | 235 | 21.0050 | 0.095 | 1.0 | 19 | 200 | 200 |
Framework versions
- Transformers 4.23.1
- Pytorch 1.12.1+cu113
- Datasets 1.18.3
- Tokenizers 0.13.2