metadata
license: mit
tags:
- generated_from_trainer
model-index:
- name: prompt-extend
results: []
Prompt Extend
GPT-2 model trained on dataset of stable diffusion prompts.
Intended uses
Extend stable diffusion prompts with suitable style cues.
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- train_batch_size: 32
- eval_batch_size: 64
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 6.3816 | 0.35 | 100 | 4.1823 |
| 3.7123 | 0.69 | 200 | 3.3033 |
| 3.118 | 1.04 | 300 | 2.8311 |
| 2.7291 | 1.39 | 400 | 2.5503 |
| 2.4918 | 1.74 | 500 | 2.3653 |
| 2.3379 | 2.08 | 600 | 2.2375 |
| 2.1952 | 2.43 | 700 | 2.1714 |
| 2.1593 | 2.78 | 800 | 2.1453 |
Framework versions
- Transformers 4.23.1
- Pytorch 1.12.1+cu113
- Datasets 2.6.1
- Tokenizers 0.13.1