|
--- |
|
library_name: peft |
|
license: llama3.2 |
|
base_model: meta-llama/Llama-3.2-3B-Instruct |
|
tags: |
|
- llama-factory |
|
- lora |
|
- generated_from_trainer |
|
model-index: |
|
- name: qlora-llama3b-all |
|
results: [] |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# qlora-llama3b-all |
|
|
|
This model is a fine-tuned version of [meta-llama/Llama-3.2-3B-Instruct](https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct) on the train-all dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.5163 |
|
|
|
## 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: 0.0002 |
|
- train_batch_size: 1 |
|
- eval_batch_size: 1 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 8 |
|
- total_train_batch_size: 8 |
|
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments |
|
- lr_scheduler_type: cosine |
|
- lr_scheduler_warmup_ratio: 0.1 |
|
- training_steps: 500 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:------:|:----:|:---------------:| |
|
| 2.2556 | 0.0889 | 10 | 1.8993 | |
|
| 1.5026 | 0.1778 | 20 | 0.8436 | |
|
| 0.8291 | 0.2667 | 30 | 0.6634 | |
|
| 0.6688 | 0.3556 | 40 | 0.5723 | |
|
| 0.6339 | 0.4444 | 50 | 0.5511 | |
|
| 0.5258 | 0.5333 | 60 | 0.4760 | |
|
| 0.4825 | 0.6222 | 70 | 0.4696 | |
|
| 0.5488 | 0.7111 | 80 | 0.4699 | |
|
| 0.4231 | 0.8 | 90 | 0.4597 | |
|
| 0.4558 | 0.8889 | 100 | 0.4219 | |
|
| 0.4588 | 0.9778 | 110 | 0.4175 | |
|
| 0.4592 | 1.0667 | 120 | 0.4288 | |
|
| 0.2996 | 1.1556 | 130 | 0.3918 | |
|
| 0.3269 | 1.2444 | 140 | 0.4183 | |
|
| 0.347 | 1.3333 | 150 | 0.4314 | |
|
| 0.3251 | 1.4222 | 160 | 0.3889 | |
|
| 0.3035 | 1.5111 | 170 | 0.3789 | |
|
| 0.3141 | 1.6 | 180 | 0.3869 | |
|
| 0.2878 | 1.6889 | 190 | 0.3910 | |
|
| 0.3063 | 1.7778 | 200 | 0.3958 | |
|
| 0.2748 | 1.8667 | 210 | 0.3819 | |
|
| 0.2725 | 1.9556 | 220 | 0.4040 | |
|
| 0.2897 | 2.0444 | 230 | 0.3928 | |
|
| 0.1813 | 2.1333 | 240 | 0.4048 | |
|
| 0.1965 | 2.2222 | 250 | 0.4036 | |
|
| 0.1751 | 2.3111 | 260 | 0.4221 | |
|
| 0.1739 | 2.4 | 270 | 0.4037 | |
|
| 0.1629 | 2.4889 | 280 | 0.4177 | |
|
| 0.1919 | 2.5778 | 290 | 0.4002 | |
|
| 0.1804 | 2.6667 | 300 | 0.4098 | |
|
| 0.1569 | 2.7556 | 310 | 0.4125 | |
|
| 0.1914 | 2.8444 | 320 | 0.4052 | |
|
| 0.144 | 2.9333 | 330 | 0.4041 | |
|
| 0.1738 | 3.0222 | 340 | 0.4221 | |
|
| 0.1087 | 3.1111 | 350 | 0.4214 | |
|
| 0.0876 | 3.2 | 360 | 0.4379 | |
|
| 0.0857 | 3.2889 | 370 | 0.4655 | |
|
| 0.0978 | 3.3778 | 380 | 0.4744 | |
|
| 0.0746 | 3.4667 | 390 | 0.4815 | |
|
| 0.0897 | 3.5556 | 400 | 0.4889 | |
|
| 0.0645 | 3.6444 | 410 | 0.4995 | |
|
| 0.0649 | 3.7333 | 420 | 0.5079 | |
|
| 0.0896 | 3.8222 | 430 | 0.5098 | |
|
| 0.0788 | 3.9111 | 440 | 0.5095 | |
|
| 0.0886 | 4.0 | 450 | 0.5105 | |
|
| 0.0471 | 4.0889 | 460 | 0.5111 | |
|
| 0.0461 | 4.1778 | 470 | 0.5152 | |
|
| 0.0607 | 4.2667 | 480 | 0.5152 | |
|
| 0.0473 | 4.3556 | 490 | 0.5192 | |
|
| 0.052 | 4.4444 | 500 | 0.5163 | |
|
|
|
|
|
### Framework versions |
|
|
|
- PEFT 0.12.0 |
|
- Transformers 4.46.1 |
|
- Pytorch 2.5.1+cu124 |
|
- Datasets 3.1.0 |
|
- Tokenizers 0.20.3 |
