metadata
tags:
- generated_from_trainer
metrics:
- wer
- bleu
model-index:
- name: geez_t5-15k
results: []
geez_t5-15k
This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.3233
- Wer: 0.2209
- Cer: 0.1381
- Bleu: 70.4059
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: 0.0005
- train_batch_size: 128
- eval_batch_size: 128
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 50
Training results
Training Loss | Epoch | Step | Validation Loss | Wer | Cer | Bleu |
---|---|---|---|---|---|---|
7.7095 | 1.0 | 145 | 7.7918 | 5.0898 | 3.9256 | 0.0023 |
7.0334 | 2.0 | 290 | 7.1199 | 5.0160 | 4.1855 | 0.0051 |
6.4831 | 3.0 | 435 | 6.6645 | 5.0207 | 3.8475 | 0.0214 |
6.1982 | 4.0 | 580 | 6.3920 | 4.5634 | 3.8489 | 0.0529 |
5.903 | 5.0 | 725 | 6.1877 | 4.5275 | 3.5050 | 0.0557 |
5.669 | 6.0 | 870 | 6.0360 | 4.9197 | 4.0028 | 0.0634 |
5.425 | 7.0 | 1015 | 5.8639 | 4.4216 | 3.7590 | 0.1208 |
5.2049 | 8.0 | 1160 | 5.7314 | 3.2761 | 2.6167 | 0.1783 |
5.0061 | 9.0 | 1305 | 5.6525 | 3.9136 | 3.2163 | 0.1433 |
4.8471 | 10.0 | 1450 | 5.5808 | 2.8054 | 2.4552 | 0.3077 |
4.6025 | 11.0 | 1595 | 5.4963 | 3.1738 | 2.8400 | 0.2473 |
4.4593 | 12.0 | 1740 | 5.4572 | 2.9939 | 2.6228 | 0.3764 |
4.3925 | 13.0 | 1885 | 5.3739 | 2.4268 | 2.0558 | 0.4943 |
4.2547 | 14.0 | 2030 | 5.3549 | 2.1811 | 1.9179 | 0.6141 |
4.2059 | 15.0 | 2175 | 5.3532 | 2.5793 | 2.2089 | 0.5485 |
4.0344 | 16.0 | 2320 | 5.3384 | 2.1161 | 1.8753 | 0.7106 |
3.8338 | 17.0 | 2465 | 5.3491 | 2.1119 | 1.9856 | 0.6538 |
3.8922 | 18.0 | 2610 | 5.3233 | 2.0402 | 1.8304 | 0.8877 |
3.6469 | 19.0 | 2755 | 5.3290 | 1.7011 | 1.4942 | 1.1830 |
2.8339 | 20.0 | 2900 | 4.1129 | 1.7063 | 1.4567 | 4.0465 |
1.4826 | 21.0 | 3045 | 2.3404 | 1.6510 | 1.4483 | 11.1205 |
0.8862 | 22.0 | 3190 | 1.6343 | 1.4432 | 1.2622 | 18.9607 |
0.603 | 23.0 | 3335 | 1.3605 | 1.1528 | 0.9975 | 27.6554 |
0.4701 | 24.0 | 3480 | 1.2962 | 1.0378 | 0.8913 | 31.5906 |
0.4302 | 25.0 | 3625 | 1.2630 | 0.8397 | 0.7215 | 38.0315 |
0.3239 | 26.0 | 3770 | 1.2441 | 0.6757 | 0.5460 | 44.0109 |
0.2679 | 27.0 | 3915 | 1.2520 | 0.6738 | 0.5478 | 44.8130 |
0.2543 | 28.0 | 4060 | 1.2496 | 0.6416 | 0.5215 | 46.1244 |
0.2113 | 29.0 | 4205 | 1.2534 | 0.5392 | 0.4282 | 50.5640 |
0.1811 | 30.0 | 4350 | 1.2870 | 0.6152 | 0.4961 | 47.6743 |
0.1676 | 31.0 | 4495 | 1.2657 | 0.5494 | 0.4411 | 50.7361 |
0.1523 | 32.0 | 4640 | 1.2986 | 0.5483 | 0.4476 | 50.8212 |
0.1468 | 33.0 | 4785 | 1.3057 | 0.4785 | 0.3744 | 54.2680 |
0.1375 | 34.0 | 4930 | 1.3025 | 0.4506 | 0.3545 | 55.8315 |
0.1259 | 35.0 | 5075 | 1.3367 | 0.4865 | 0.3899 | 54.1053 |
0.1194 | 36.0 | 5220 | 1.3196 | 0.4540 | 0.3581 | 55.4216 |
0.1116 | 37.0 | 5365 | 1.3104 | 0.3943 | 0.3011 | 58.6213 |
0.0968 | 38.0 | 5510 | 1.3477 | 0.3834 | 0.2953 | 59.3219 |
0.0981 | 39.0 | 5655 | 1.3217 | 0.4059 | 0.3112 | 58.2604 |
0.0938 | 40.0 | 5800 | 1.3304 | 0.4132 | 0.3205 | 57.7388 |
0.0823 | 41.0 | 5945 | 1.3023 | 0.3432 | 0.2481 | 61.8713 |
0.0786 | 42.0 | 6090 | 1.3138 | 0.2974 | 0.2027 | 64.6092 |
0.0766 | 43.0 | 6235 | 1.3324 | 0.3680 | 0.2768 | 60.6454 |
0.0765 | 44.0 | 6380 | 1.3266 | 0.3359 | 0.2359 | 62.7278 |
0.0718 | 45.0 | 6525 | 1.3440 | 0.3000 | 0.2163 | 64.6481 |
0.0637 | 46.0 | 6670 | 1.3283 | 0.2628 | 0.1782 | 67.2375 |
0.0658 | 47.0 | 6815 | 1.3331 | 0.2605 | 0.1721 | 67.1960 |
0.0643 | 48.0 | 6960 | 1.3198 | 0.2618 | 0.1780 | 67.4730 |
0.0682 | 49.0 | 7105 | 1.3196 | 0.2732 | 0.1876 | 66.2931 |
0.0605 | 50.0 | 7250 | 1.3233 | 0.2209 | 0.1381 | 70.4059 |
Framework versions
- Transformers 4.38.2
- Pytorch 2.2.1+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2