metadata
license: gemma
base_model: google/gemma-2b
tags:
- generated_from_trainer
model-index:
- name: G0428B1
results: []
G0428B1
This model is a fine-tuned version of google/gemma-2b on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.1474
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: 60
- num_epochs: 1
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.238 | 0.09 | 10 | 1.8587 |
| 1.7567 | 0.18 | 20 | 1.5411 |
| 1.2688 | 0.27 | 30 | 0.8329 |
| 0.5198 | 0.36 | 40 | 0.2499 |
| 0.1877 | 0.45 | 50 | 0.1580 |
| 0.1639 | 0.54 | 60 | 0.1526 |
| 0.1475 | 0.63 | 70 | 0.1475 |
| 0.1626 | 0.73 | 80 | 0.1470 |
| 0.1406 | 0.82 | 90 | 0.1481 |
| 0.1536 | 0.91 | 100 | 0.1477 |
| 0.1551 | 1.0 | 110 | 0.1474 |
Framework versions
- Transformers 4.36.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1