TinyLlama2-jp-122M-FlashAttention2
This model is a fine-tuned version of TinyLlama/TinyLlama-1.1B-Chat-v1.0 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 3.3043
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: 156
- eval_batch_size: 156
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 200
- num_epochs: 1
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 7.4854 | 0.06 | 100 | 5.4181 |
| 4.9365 | 0.12 | 200 | 4.6000 |
| 4.4045 | 0.18 | 300 | 4.2480 |
| 4.1233 | 0.24 | 400 | 4.0493 |
| 3.9441 | 0.29 | 500 | 3.9019 |
| 3.8179 | 0.35 | 600 | 3.7969 |
| 3.7142 | 0.41 | 700 | 3.7099 |
| 3.6299 | 0.47 | 800 | 3.6369 |
| 3.5446 | 0.53 | 900 | 3.5616 |
| 3.4792 | 0.59 | 1000 | 3.5006 |
| 3.4116 | 0.65 | 1100 | 3.4380 |
| 3.362 | 0.71 | 1200 | 3.3931 |
| 3.3219 | 0.77 | 1300 | 3.3552 |
| 3.2822 | 0.82 | 1400 | 3.3295 |
| 3.2567 | 0.88 | 1500 | 3.3126 |
| 3.2592 | 0.94 | 1600 | 3.3043 |
Framework versions
- Transformers 4.38.2
- Pytorch 2.1.0+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2