Text-to-image finetuning - yeonsikc/unet-sd15-conist300k
This pipeline was finetuned from runwayml/stable-diffusion-v1-5 on the None dataset. Below are some example images generated with the finetuned pipeline using the following prompts: ['masterpiece, best quality, 1girl with long white hair sitting in a field of green plants and flowers, her hand under her chin, warm lighting, white dress, blurry foreground', 'masterpiece, best quality,1girl with long white hair sitting in a field of green plants and flowers, her hand under her chin, warm lighting, white dress, blurry foreground', '1girl lying, in a field of flowers, red dress, white flower, earrings, looking at viewer, blue eyes, red ribbon, necklace, long sleeves, blonde hair, short hair, daisy', 'black border, dynamic angle, masterpiece, best quality, 1girl, solo focus ,school uniform, black hair, upper body, short hair, serafuku, train station, school bag, crowd, depth of field', 'foreshortening, depth of field, masterpiece, best quality, 1girl, brown hair, brown eyes, long hair, underwater, air bubble, solo, looking at viewer, school swimsuit, swimming, dappled sunlight', 'foreshortening, depth of field, masterpiece, best quality, 1girl, brown hair, brown eyes, long hair, underwater, air bubble, solo, looking at viewer, school swimsuit, swimming, dappled sunlight', 'masterpiece, best quality, 1boy with short white hair sitting in a field of green plants and flowers, his hand under his chin, warm lighting, white shirt, blurry foreground', 'masterpiece, best quality,1boy with short white hair sitting in a field of green plants and flowers, his hand under his chin, warm lighting, purple t-shirt, blurry foreground', '1boy lying, in a field of flowers, red shirt, white flower, earrings, looking at viewer, blue eyes, red ribbon, necklace, long sleeves, blonde hair, short hair, daisy', 'black border, dynamic angle, masterpiece, best quality, 1boy, solo focus ,school uniform, black hair, upper body, short hair, serafuku, train station, school bag, crowd, depth of field', 'foreshortening, depth of field, masterpiece, best quality, 1boy, brown hair, brown eyes, long hair, underwater, air bubble, solo, looking at viewer, school swimsuit, swimming, dappled sunlight', 'foreshortening, depth of field, masterpiece, best quality, 1boy, brown hair, brown eyes, long hair, underwater, air bubble, solo, looking at viewer, school swimsuit, swimming, dappled sunlight']:
Pipeline usage
You can use the pipeline like so:
from diffusers import DiffusionPipeline
import torch
pipeline = DiffusionPipeline.from_pretrained("yeonsikc/unet-sd15-conist300k", torch_dtype=torch.float16)
prompt = "masterpiece, best quality, 1girl with long white hair sitting in a field of green plants and flowers, her hand under her chin, warm lighting, white dress, blurry foreground"
image = pipeline(prompt).images[0]
image.save("my_image.png")
Training info
These are the key hyperparameters used during training:
- Epochs: 120
- Learning rate: 5e-05
- Batch size: 4
- Gradient accumulation steps: 1
- Image resolution: 512
- Mixed-precision: fp16
More information on all the CLI arguments and the environment are available on your wandb
run page.