metadata
base_model: Bllossom/llama-3.2-Korean-Bllossom-AICA-5B
library_name: peft
license: llama3.2
tags:
- generated_from_trainer
model-index:
- name: HGU_rulebook-Llama3.2-Bllossom-5B_fine-tuning-QLoRA-32_64_3
results: []
HGU_rulebook-Llama3.2-Bllossom-5B_fine-tuning-QLoRA-32_64_3
This model is a fine-tuned version of Bllossom/llama-3.2-Korean-Bllossom-AICA-5B on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 5.6952
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: 1e-05
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 16
- optimizer: Use adamw_bnb_8bit with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine_with_restarts
- lr_scheduler_warmup_ratio: 0.1
- training_steps: 942
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 12.459 | 0.2991 | 47 | 11.8428 |
| 8.9527 | 0.5982 | 94 | 8.3608 |
| 6.8438 | 0.8974 | 141 | 6.6652 |
| 6.1272 | 1.1965 | 188 | 6.0176 |
| 5.858 | 1.4956 | 235 | 5.8316 |
| 5.7707 | 1.7947 | 282 | 5.7618 |
| 5.7435 | 2.0939 | 329 | 5.7331 |
| 5.7215 | 2.3930 | 376 | 5.7194 |
| 5.7118 | 2.6921 | 423 | 5.7114 |
| 5.7047 | 2.9912 | 470 | 5.7063 |
| 5.703 | 3.2904 | 517 | 5.7027 |
| 5.6973 | 3.5895 | 564 | 5.7001 |
| 5.6959 | 3.8886 | 611 | 5.6985 |
| 5.6921 | 4.1877 | 658 | 5.6972 |
| 5.6968 | 4.4869 | 705 | 5.6965 |
| 5.6954 | 4.7860 | 752 | 5.6959 |
| 5.6958 | 5.0851 | 799 | 5.6955 |
| 5.6937 | 5.3842 | 846 | 5.6953 |
| 5.6915 | 5.6834 | 893 | 5.6952 |
| 5.6951 | 5.9825 | 940 | 5.6952 |
Framework versions
- PEFT 0.12.0
- Transformers 4.46.2
- Pytorch 2.0.1+cu118
- Datasets 3.0.0
- Tokenizers 0.20.1