tinyllama_mole_sft_ultrachat_ep3
This model was trained from scratch on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 1.1127
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: 2e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- total_eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 120
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.3007 | 0.09 | 100 | 1.2780 |
| 1.2255 | 0.18 | 200 | 1.2158 |
| 1.192 | 0.26 | 300 | 1.1921 |
| 1.1696 | 0.35 | 400 | 1.1770 |
| 1.1426 | 0.44 | 500 | 1.1666 |
| 1.1628 | 0.53 | 600 | 1.1583 |
| 1.1501 | 0.61 | 700 | 1.1513 |
| 1.137 | 0.7 | 800 | 1.1457 |
| 1.1321 | 0.79 | 900 | 1.1407 |
| 1.1156 | 0.88 | 1000 | 1.1359 |
| 1.1395 | 0.96 | 1100 | 1.1318 |
| 1.0564 | 1.05 | 1200 | 1.1315 |
| 1.0594 | 1.14 | 1300 | 1.1295 |
| 1.0711 | 1.23 | 1400 | 1.1274 |
| 1.0624 | 1.31 | 1500 | 1.1256 |
| 1.0652 | 1.4 | 1600 | 1.1233 |
| 1.0626 | 1.49 | 1700 | 1.1213 |
| 1.0457 | 1.58 | 1800 | 1.1195 |
| 1.0665 | 1.66 | 1900 | 1.1178 |
| 1.07 | 1.75 | 2000 | 1.1158 |
| 1.0567 | 1.84 | 2100 | 1.1141 |
| 1.0304 | 1.93 | 2200 | 1.1127 |
| 1.0132 | 2.01 | 2300 | 1.1170 |
| 1.0203 | 2.1 | 2400 | 1.1170 |
| 1.0088 | 2.19 | 2500 | 1.1168 |
| 1.002 | 2.28 | 2600 | 1.1162 |
| 1.0004 | 2.37 | 2700 | 1.1157 |
| 1.0058 | 2.45 | 2800 | 1.1156 |
| 1.0118 | 2.54 | 2900 | 1.1150 |
| 0.9941 | 2.63 | 3000 | 1.1148 |
| 1.0127 | 2.72 | 3100 | 1.1147 |
| 1.0039 | 2.8 | 3200 | 1.1144 |
| 1.0 | 2.89 | 3300 | 1.1143 |
| 1.0188 | 2.98 | 3400 | 1.1143 |
Framework versions
- Transformers 4.37.0
- Pytorch 2.1.2+cu118
- Datasets 2.16.1
- Tokenizers 0.15.0