limbxy_seq_t2_heads4_layers2
This model is a fine-tuned version of c14kevincardenas/beit-large-patch16-384-limb on the c14kevincardenas/beta_caller_284_limbxy_seq_2 dataset. It achieves the following results on the evaluation set:
- Loss: 0.0053
- Rmse: 0.0727
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: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 2014
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 250
- num_epochs: 50.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rmse |
|---|---|---|---|---|
| 0.0146 | 1.0 | 150 | 0.0181 | 0.1345 |
| 0.011 | 2.0 | 300 | 0.0086 | 0.0930 |
| 0.0089 | 3.0 | 450 | 0.0080 | 0.0892 |
| 0.0058 | 4.0 | 600 | 0.0086 | 0.0929 |
| 0.0041 | 5.0 | 750 | 0.0089 | 0.0941 |
| 0.003 | 6.0 | 900 | 0.0063 | 0.0791 |
| 0.003 | 7.0 | 1050 | 0.0070 | 0.0836 |
| 0.0026 | 8.0 | 1200 | 0.0074 | 0.0859 |
| 0.002 | 9.0 | 1350 | 0.0063 | 0.0796 |
| 0.0021 | 10.0 | 1500 | 0.0062 | 0.0788 |
| 0.0018 | 11.0 | 1650 | 0.0066 | 0.0814 |
| 0.002 | 12.0 | 1800 | 0.0058 | 0.0759 |
| 0.0011 | 13.0 | 1950 | 0.0063 | 0.0795 |
| 0.0012 | 14.0 | 2100 | 0.0058 | 0.0763 |
| 0.0012 | 15.0 | 2250 | 0.0059 | 0.0770 |
| 0.0009 | 16.0 | 2400 | 0.0057 | 0.0756 |
| 0.001 | 17.0 | 2550 | 0.0058 | 0.0760 |
| 0.0008 | 18.0 | 2700 | 0.0056 | 0.0745 |
| 0.0008 | 19.0 | 2850 | 0.0058 | 0.0761 |
| 0.0006 | 20.0 | 3000 | 0.0057 | 0.0757 |
| 0.0007 | 21.0 | 3150 | 0.0056 | 0.0750 |
| 0.0005 | 22.0 | 3300 | 0.0057 | 0.0758 |
| 0.0005 | 23.0 | 3450 | 0.0056 | 0.0749 |
| 0.0006 | 24.0 | 3600 | 0.0054 | 0.0735 |
| 0.0005 | 25.0 | 3750 | 0.0060 | 0.0772 |
| 0.0005 | 26.0 | 3900 | 0.0057 | 0.0755 |
| 0.0005 | 27.0 | 4050 | 0.0056 | 0.0748 |
| 0.0006 | 28.0 | 4200 | 0.0055 | 0.0738 |
| 0.0004 | 29.0 | 4350 | 0.0056 | 0.0751 |
| 0.0004 | 30.0 | 4500 | 0.0054 | 0.0734 |
| 0.0004 | 31.0 | 4650 | 0.0053 | 0.0730 |
| 0.0003 | 32.0 | 4800 | 0.0053 | 0.0728 |
| 0.0003 | 33.0 | 4950 | 0.0054 | 0.0732 |
| 0.0003 | 34.0 | 5100 | 0.0054 | 0.0732 |
| 0.0003 | 35.0 | 5250 | 0.0053 | 0.0727 |
| 0.0002 | 36.0 | 5400 | 0.0054 | 0.0734 |
| 0.0003 | 37.0 | 5550 | 0.0054 | 0.0735 |
| 0.0002 | 38.0 | 5700 | 0.0054 | 0.0735 |
| 0.0002 | 39.0 | 5850 | 0.0055 | 0.0742 |
| 0.0002 | 40.0 | 6000 | 0.0054 | 0.0736 |
| 0.0002 | 41.0 | 6150 | 0.0054 | 0.0732 |
| 0.0001 | 42.0 | 6300 | 0.0053 | 0.0727 |
| 0.0001 | 43.0 | 6450 | 0.0054 | 0.0734 |
| 0.0001 | 44.0 | 6600 | 0.0054 | 0.0733 |
| 0.0001 | 45.0 | 6750 | 0.0053 | 0.0730 |
| 0.0001 | 46.0 | 6900 | 0.0053 | 0.0729 |
| 0.0001 | 47.0 | 7050 | 0.0053 | 0.0731 |
| 0.0001 | 48.0 | 7200 | 0.0053 | 0.0730 |
| 0.0001 | 49.0 | 7350 | 0.0053 | 0.0730 |
| 0.0001 | 50.0 | 7500 | 0.0053 | 0.0731 |
Framework versions
- Transformers 4.45.2
- Pytorch 2.5.0+cu124
- Datasets 3.0.1
- Tokenizers 0.20.1