metadata
license: other
base_model: yahma/llama-7b-hf
tags:
- generated_from_trainer
model-index:
- name: V0224B2
results: []
V0224B2
This model is a fine-tuned version of yahma/llama-7b-hf on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.7905
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 |
|---|---|---|---|
| 1.116 | 0.13 | 10 | 1.1293 |
| 1.1164 | 0.26 | 20 | 1.1077 |
| 1.0694 | 0.39 | 30 | 1.0223 |
| 0.971 | 0.52 | 40 | 0.9344 |
| 0.9108 | 0.65 | 50 | 0.8913 |
| 0.873 | 0.78 | 60 | 0.8662 |
| 0.8345 | 0.91 | 70 | 0.8457 |
| 0.8232 | 1.04 | 80 | 0.8333 |
| 0.8028 | 1.17 | 90 | 0.8228 |
| 0.8148 | 1.3 | 100 | 0.8154 |
| 0.8001 | 1.43 | 110 | 0.8098 |
| 0.7887 | 1.55 | 120 | 0.8050 |
| 0.7757 | 1.68 | 130 | 0.8015 |
| 0.7872 | 1.81 | 140 | 0.7988 |
| 0.7928 | 1.94 | 150 | 0.7962 |
| 0.7754 | 2.07 | 160 | 0.7941 |
| 0.7641 | 2.2 | 170 | 0.7928 |
| 0.7758 | 2.33 | 180 | 0.7917 |
| 0.768 | 2.46 | 190 | 0.7912 |
| 0.7691 | 2.59 | 200 | 0.7909 |
| 0.7728 | 2.72 | 210 | 0.7906 |
| 0.769 | 2.85 | 220 | 0.7904 |
| 0.7715 | 2.98 | 230 | 0.7905 |
Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1