|
--- |
|
license: other |
|
base_model: yahma/llama-7b-hf |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: V0305B2 |
|
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. --> |
|
|
|
# V0305B2 |
|
|
|
This model is a fine-tuned version of [yahma/llama-7b-hf](https://huggingface.co./yahma/llama-7b-hf) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.0894 |
|
|
|
## 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: 0.0003 |
|
- train_batch_size: 4 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 32 |
|
- total_train_batch_size: 128 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: cosine_with_restarts |
|
- lr_scheduler_warmup_steps: 20 |
|
- num_epochs: 3 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 2.352 | 0.09 | 10 | 2.3256 | |
|
| 2.1754 | 0.17 | 20 | 1.8064 | |
|
| 1.2745 | 0.26 | 30 | 0.6844 | |
|
| 0.3789 | 0.34 | 40 | 0.1687 | |
|
| 0.1587 | 0.43 | 50 | 0.1487 | |
|
| 0.1563 | 0.51 | 60 | 0.1506 | |
|
| 0.1505 | 0.6 | 70 | 0.1502 | |
|
| 0.1525 | 0.68 | 80 | 0.1487 | |
|
| 0.1481 | 0.77 | 90 | 0.1492 | |
|
| 0.1504 | 0.85 | 100 | 0.1441 | |
|
| 0.1501 | 0.94 | 110 | 0.1436 | |
|
| 0.1439 | 1.02 | 120 | 0.1360 | |
|
| 0.1411 | 1.11 | 130 | 0.1276 | |
|
| 0.1349 | 1.19 | 140 | 0.1259 | |
|
| 0.1345 | 1.28 | 150 | 0.1190 | |
|
| 0.1299 | 1.37 | 160 | 0.1114 | |
|
| 0.1275 | 1.45 | 170 | 0.1058 | |
|
| 0.1159 | 1.54 | 180 | 0.1013 | |
|
| 0.1189 | 1.62 | 190 | 0.0997 | |
|
| 0.1203 | 1.71 | 200 | 0.1012 | |
|
| 0.1177 | 1.79 | 210 | 0.0973 | |
|
| 0.1144 | 1.88 | 220 | 0.0932 | |
|
| 0.1128 | 1.96 | 230 | 0.0933 | |
|
| 0.1084 | 2.05 | 240 | 0.0952 | |
|
| 0.1081 | 2.13 | 250 | 0.0930 | |
|
| 0.1037 | 2.22 | 260 | 0.0921 | |
|
| 0.1011 | 2.3 | 270 | 0.0923 | |
|
| 0.1072 | 2.39 | 280 | 0.0912 | |
|
| 0.1058 | 2.47 | 290 | 0.0902 | |
|
| 0.1107 | 2.56 | 300 | 0.0899 | |
|
| 0.1066 | 2.65 | 310 | 0.0897 | |
|
| 0.1091 | 2.73 | 320 | 0.0895 | |
|
| 0.103 | 2.82 | 330 | 0.0893 | |
|
| 0.1021 | 2.9 | 340 | 0.0893 | |
|
| 0.103 | 2.99 | 350 | 0.0894 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.36.0.dev0 |
|
- Pytorch 2.1.2+cu121 |
|
- Datasets 2.14.6 |
|
- Tokenizers 0.14.1 |
|
|
