metadata
license: other
library_name: peft
tags:
- generated_from_trainer
base_model: beomi/Llama-3-Open-Ko-8B-Instruct-preview
model-index:
- name: llama8_on_korean_summary
results: []
llama8_on_korean_summary
This model is a fine-tuned version of beomi/Llama-3-Open-Ko-8B-Instruct-preview on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.8512
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: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 5
- total_train_batch_size: 10
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- training_steps: 500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.5836 | 0.26 | 20 | 1.4113 |
| 1.2717 | 0.53 | 40 | 1.1577 |
| 1.1128 | 0.79 | 60 | 1.0624 |
| 1.0331 | 1.05 | 80 | 1.0005 |
| 0.9565 | 1.32 | 100 | 0.9578 |
| 0.9084 | 1.58 | 120 | 0.9194 |
| 0.9038 | 1.84 | 140 | 0.8943 |
| 0.8157 | 2.11 | 160 | 0.8798 |
| 0.7401 | 2.37 | 180 | 0.8598 |
| 0.7218 | 2.63 | 200 | 0.8412 |
| 0.725 | 2.89 | 220 | 0.8264 |
| 0.6651 | 3.16 | 240 | 0.8289 |
| 0.5981 | 3.42 | 260 | 0.8238 |
| 0.5987 | 3.68 | 280 | 0.8149 |
| 0.5998 | 3.95 | 300 | 0.8093 |
| 0.5423 | 4.21 | 320 | 0.8248 |
| 0.5188 | 4.47 | 340 | 0.8221 |
| 0.5258 | 4.74 | 360 | 0.8201 |
| 0.5028 | 5.0 | 380 | 0.8199 |
| 0.4567 | 5.26 | 400 | 0.8421 |
| 0.4558 | 5.53 | 420 | 0.8427 |
| 0.4424 | 5.79 | 440 | 0.8395 |
| 0.4456 | 6.05 | 460 | 0.8387 |
| 0.4147 | 6.32 | 480 | 0.8517 |
| 0.4254 | 6.58 | 500 | 0.8512 |
Framework versions
- PEFT 0.8.2
- Transformers 4.38.2
- Pytorch 2.1.2+cu121
- Datasets 2.17.0
- Tokenizers 0.15.0