gpt-neo-125M-magicprompt-SD

Generate/augment your prompt, stable diffusion style.

This model is a fine-tuned version of EleutherAI/gpt-neo-125M on the Gustavosta/Stable-Diffusion-Prompts dataset. It achieves the following results on the evaluation set:

• Loss: 1.8875
• perplexity: 6.6028

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: 16
• eval_batch_size: 2
• seed: 42
• distributed_type: multi-GPU
• num_devices: 2
• gradient_accumulation_steps: 8
• total_train_batch_size: 256
• 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
3.2189	0.99	33	3.0051
2.5466	1.99	66	2.5215
2.2791	2.99	99	2.2881
2.107	3.99	132	2.1322
1.9458	4.99	165	2.0270
1.8664	5.99	198	1.9580
1.8083	6.99	231	1.9177
1.7631	7.99	264	1.8964
1.7369	8.99	297	1.8885
1.766	9.99	330	1.8875

Framework versions

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