Baby-Llama-58M-ORIGINAL
This model is a fine-tuned version of on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 4.1715
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: 0.00025
- train_batch_size: 32
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 50
- num_epochs: 80
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 289.4006 | 1.0 | 12 | 243.1602 |
| 229.6677 | 2.0 | 24 | 201.2611 |
| 207.9305 | 3.0 | 36 | 172.3865 |
| 125.9577 | 4.0 | 48 | 107.6148 |
| 84.8722 | 5.0 | 60 | 72.6301 |
| 48.2019 | 6.0 | 72 | 40.8081 |
| 25.796 | 7.0 | 84 | 22.0368 |
| 15.6686 | 8.0 | 96 | 13.6649 |
| 9.8745 | 9.0 | 108 | 9.6135 |
| 7.9539 | 10.0 | 120 | 8.0438 |
| 6.4873 | 11.0 | 132 | 7.1011 |
| 6.0896 | 12.0 | 144 | 6.4537 |
| 5.4521 | 13.0 | 156 | 6.0605 |
| 5.5516 | 14.0 | 168 | 6.0324 |
| 4.7538 | 15.0 | 180 | 5.7866 |
| 4.8229 | 16.0 | 192 | 5.5738 |
| 4.568 | 17.0 | 204 | 5.5282 |
| 4.4449 | 18.0 | 216 | 5.4060 |
| 4.6567 | 19.0 | 228 | 5.3382 |
| 4.1888 | 20.0 | 240 | 5.2407 |
| 4.2102 | 21.0 | 252 | 5.2085 |
| 4.6584 | 22.0 | 264 | 5.0947 |
| 4.102 | 23.0 | 276 | 4.9988 |
| 4.3574 | 24.0 | 288 | 4.9768 |
| 4.0571 | 25.0 | 300 | 4.9552 |
| 4.22 | 26.0 | 312 | 4.9127 |
| 3.9908 | 27.0 | 324 | 4.9050 |
| 4.0273 | 28.0 | 336 | 4.7905 |
| 4.0092 | 29.0 | 348 | 4.8265 |
| 3.9705 | 30.0 | 360 | 4.7823 |
| 4.0081 | 31.0 | 372 | 4.7383 |
| 3.8771 | 32.0 | 384 | 4.6774 |
| 3.899 | 33.0 | 396 | 4.6629 |
| 3.4711 | 34.0 | 408 | 4.6603 |
| 3.4489 | 35.0 | 420 | 4.5675 |
| 3.5063 | 36.0 | 432 | 4.5751 |
| 3.6348 | 37.0 | 444 | 4.5786 |
| 3.6931 | 38.0 | 456 | 4.5513 |
| 3.7022 | 39.0 | 468 | 4.5208 |
| 3.6842 | 40.0 | 480 | 4.5146 |
| 3.4084 | 41.0 | 492 | 4.5171 |
| 3.5141 | 42.0 | 504 | 4.4681 |
| 3.2337 | 43.0 | 516 | 4.4700 |
| 3.4376 | 44.0 | 528 | 4.4472 |
| 3.2911 | 45.0 | 540 | 4.4462 |
| 3.6011 | 46.0 | 552 | 4.4115 |
| 3.8547 | 47.0 | 564 | 4.3901 |
| 3.3866 | 48.0 | 576 | 4.3873 |
| 3.4543 | 49.0 | 588 | 4.3904 |
| 3.6357 | 50.0 | 600 | 4.3693 |
| 3.5045 | 51.0 | 612 | 4.3569 |
| 3.0792 | 52.0 | 624 | 4.3263 |
| 3.2731 | 53.0 | 636 | 4.3322 |
| 3.4193 | 54.0 | 648 | 4.3012 |
| 3.1097 | 55.0 | 660 | 4.3015 |
| 3.088 | 56.0 | 672 | 4.2914 |
| 2.9444 | 57.0 | 684 | 4.2750 |
| 3.362 | 58.0 | 696 | 4.2612 |
| 3.2228 | 59.0 | 708 | 4.2647 |
| 2.9892 | 60.0 | 720 | 4.2417 |
| 3.0214 | 61.0 | 732 | 4.2287 |
| 3.3049 | 62.0 | 744 | 4.2328 |
| 3.4639 | 63.0 | 756 | 4.2200 |
| 3.2505 | 64.0 | 768 | 4.2130 |
| 3.0121 | 65.0 | 780 | 4.2087 |
| 3.3112 | 66.0 | 792 | 4.2001 |
| 3.3258 | 67.0 | 804 | 4.2013 |
| 2.9143 | 68.0 | 816 | 4.1952 |
| 3.1404 | 69.0 | 828 | 4.1876 |
| 3.495 | 70.0 | 840 | 4.1910 |
| 3.134 | 71.0 | 852 | 4.1841 |
| 3.1945 | 72.0 | 864 | 4.1835 |
| 3.1116 | 73.0 | 876 | 4.1742 |
| 3.2141 | 74.0 | 888 | 4.1743 |
| 3.3962 | 75.0 | 900 | 4.1734 |
| 2.8472 | 76.0 | 912 | 4.1721 |
| 3.2455 | 77.0 | 924 | 4.1717 |
| 2.9047 | 78.0 | 936 | 4.1724 |
| 3.4182 | 79.0 | 948 | 4.1716 |
| 3.5737 | 80.0 | 960 | 4.1715 |
Framework versions
- Transformers 4.39.1
- Pytorch 2.1.2+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0