theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-04-24 23:26:15 +00:00
Code Issues Packages Projects Releases Wiki Activity

add full training

Browse Source
This commit is contained in:
mi804
2026-04-24 15:11:34 +08:00
parent 3799bdc23a
commit 2d7d5137ea
4 changed files with 85 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
examples/stable_diffusion/model_training/full/stable-diffusion-v1-5.sh Normal file
View File

@@ -0,0 +1,15 @@
modelscope download --dataset DiffSynth-Studio/diffsynth_example_dataset --include "stable_diffusion/stable-diffusion-v1-5/*" --local_dir ./data/diffsynth_example_dataset
accelerate launch examples/stable_diffusion/model_training/train.py \
--dataset_base_path data/diffsynth_example_dataset/stable_diffusion/stable-diffusion-v1-5 \
--dataset_metadata_path data/diffsynth_example_dataset/stable_diffusion/stable-diffusion-v1-5/metadata.csv \
--height 512 \
--width 512 \
--dataset_repeat 50 \
--model_id_with_origin_paths "AI-ModelScope/stable-diffusion-v1-5:text_encoder/model.safetensors,AI-ModelScope/stable-diffusion-v1-5:unet/diffusion_pytorch_model.safetensors,AI-ModelScope/stable-diffusion-v1-5:vae/diffusion_pytorch_model.safetensors" \
--learning_rate 1e-5 \
--num_epochs 2 \
--trainable_models "unet" \
--remove_prefix_in_ckpt "pipe.unet." \
--output_path "./models/train/stable-diffusion-v1-5_full" \
--use_gradient_checkpointing

27
examples/stable_diffusion/model_training/validate_full/stable-diffusion-v1-5.py Normal file
View File

@@ -0,0 +1,27 @@
from diffsynth.pipelines.stable_diffusion import StableDiffusionPipeline, ModelConfig
from diffsynth.core import load_state_dict
import torch
pipe = StableDiffusionPipeline.from_pretrained(
torch_dtype=torch.float32,
model_configs=[
ModelConfig(model_id="AI-ModelScope/stable-diffusion-v1-5", origin_file_pattern="text_encoder/model.safetensors"),
ModelConfig(model_id="AI-ModelScope/stable-diffusion-v1-5", origin_file_pattern="unet/diffusion_pytorch_model.safetensors"),
ModelConfig(model_id="AI-ModelScope/stable-diffusion-v1-5", origin_file_pattern="vae/diffusion_pytorch_model.safetensors"),
],
tokenizer_config=ModelConfig(model_id="AI-ModelScope/stable-diffusion-v1-5", origin_file_pattern="tokenizer/"),
)
state_dict = load_state_dict("./models/train/stable-diffusion-v1-5_full/epoch-1.safetensors", torch_dtype=torch.float32)
pipe.unet.load_state_dict(state_dict)
image = pipe(
prompt="a dog",
negative_prompt="blurry, low quality, deformed",
cfg_scale=7.5,
height=512,
width=512,
seed=42,
rand_device="cuda",
num_inference_steps=50,
)
image.save("image_stable-diffusion-v1-5_full.jpg")