metadata
license: other
library_name: peft
tags:
- trl
- sft
- generated_from_trainer
base_model: google/gemma-2b
datasets:
- generator
model-index:
- name: gemma2bit-sql
results: []
gemma2bit-sql
This model is a fine-tuned version of google/gemma-2b on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 0.7133
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: 4
- eval_batch_size: 8
- seed: 1399
- gradient_accumulation_steps: 8
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_steps: 10
- training_steps: 100
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.7834 | 0.02 | 5 | 1.5323 |
| 1.383 | 0.04 | 10 | 1.1950 |
| 1.1294 | 0.06 | 15 | 1.0211 |
| 0.964 | 0.08 | 20 | 0.9335 |
| 0.9112 | 0.09 | 25 | 0.8822 |
| 0.8775 | 0.11 | 30 | 0.8455 |
| 0.8375 | 0.13 | 35 | 0.8225 |
| 0.8109 | 0.15 | 40 | 0.8011 |
| 0.783 | 0.17 | 45 | 0.7862 |
| 0.7784 | 0.19 | 50 | 0.7738 |
| 0.7698 | 0.21 | 55 | 0.7625 |
| 0.7627 | 0.23 | 60 | 0.7524 |
| 0.7599 | 0.24 | 65 | 0.7461 |
| 0.7482 | 0.26 | 70 | 0.7406 |
| 0.7368 | 0.28 | 75 | 0.7338 |
| 0.7407 | 0.3 | 80 | 0.7293 |
| 0.7211 | 0.32 | 85 | 0.7243 |
| 0.718 | 0.34 | 90 | 0.7214 |
| 0.7187 | 0.36 | 95 | 0.7167 |
| 0.7251 | 0.38 | 100 | 0.7133 |
Framework versions
- PEFT 0.9.0
- Transformers 4.38.2
- Pytorch 2.1.0+cu118
- Datasets 2.18.0
- Tokenizers 0.15.2