tiny-gpt2-magicprompt

Generate/augment your prompt, stable diffusion style. Enter a new dimension of creativity

This model is a fine-tuned version of sshleifer/tiny-gpt2 on the Gustavosta/Stable-Diffusion-Prompts dataset. It achieves the following results on the evaluation set:

• Loss: 10.7918
• perplexity: 48618.8756

Intended uses & limitations

???

Training and evaluation data

refer to the Gustavosta/Stable-Diffusion-Prompts dataset.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0001
• train_batch_size: 8
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 32
• total_train_batch_size: 512
• total_eval_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss
10.8201	0.96	16	10.8191
10.8167	1.96	32	10.8145
10.8117	2.96	48	10.8095
10.8058	3.96	64	10.8025
10.7997	4.96	80	10.7989
10.7959	5.96	96	10.7947
10.7934	6.96	112	10.7925
10.7924	7.96	128	10.7919
10.7921	8.96	144	10.7918
10.792	9.96	160	10.7918

Framework versions

• Transformers 4.25.0.dev0
• Pytorch 1.13.0+cu117
• Datasets 2.6.1
• Tokenizers 0.13.1