本文主要为Datawhale暑期夏令营第四期AIGC方向Task01学习笔记,官方学习连接:Datawhale夏令营第四期AIGC方向Task01
修行成果:
1 赛事内容
本次比赛为阿里云天池举办的可图Kolors-LoRA风格故事挑战赛(比赛链接),具体赛事任务如下:
1.参赛者需在可图Kolors 模型的基础上训练LoRA 模型,生成无限风格,如水墨画风格、水彩风格、赛博朋克风格、日漫风格…
2.基于LoRA模型生成 8 张图片组成连贯故事,故事内容可自定义;基于8图故事,评估LoRA风格的美感度及连贯性
样例:偶像少女养成日记
2 baseline.ipynb代码注解
2.1 基本库的安装
!pip install simple-aesthetics-predictor
!pip install -v -e data-juicer
!pip uninstall pytorch-lightning -y
!pip install peft lightning pandas torchvision
!pip install -e DiffSynth-Studio
!pip install对一些库进行安装,实现对环境的基本配置,以便后续程序运行
2.2 数据集及其处理
from modelscope.msdatasets import MsDataset
ds = MsDataset.load(
'AI-ModelScope/lowres_anime',
subset_name='default',
split='train',
cache_dir="/mnt/workspace/kolors/data"
)
import json, os
from data_juicer.utils.mm_utils import SpecialTokens
from tqdm import tqdm
os.makedirs("./data/lora_dataset/train", exist_ok=True)
os.makedirs("./data/data-juicer/input", exist_ok=True)
with open("./data/data-juicer/input/metadata.jsonl", "w") as f:
for data_id, data in enumerate(tqdm(ds)):
image = data["image"].convert("RGB")
image.save(f"/mnt/workspace/kolors/data/lora_dataset/train/{data_id}.jpg")
metadata = {"text": "二次元", "image": [f"/mnt/workspace/kolors/data/lora_dataset/train/{data_id}.jpg"]}
f.write(json.dumps(metadata))
f.write("\n")
此两段代码下载并保存数据集中的图片及元数据
2.3 使用 data-juicer 处理数据
data_juicer_config = """
# global parameters
project_name: 'data-process'
dataset_path: './data/data-juicer/input/metadata.jsonl' # path to your dataset directory or file
np: 4 # number of subprocess to process your dataset
text_keys: 'text'
image_key: 'image'
image_special_token: '<__dj__image>'
export_path: './data/data-juicer/output/result.jsonl'
# process schedule
# a list of several process operators with their arguments
process:
- image_shape_filter:
min_width: 1024
min_height: 1024
any_or_all: any
- image_aspect_ratio_filter:
min_ratio: 0.5
max_ratio: 2.0
any_or_all: any
"""
with open("data/data-juicer/data_juicer_config.yaml", "w") as file:
file.write(data_juicer_config.strip())
!dj-process --config data/data-juicer/data_juicer_config.yaml
保存处理好的数据
import pandas as pd
import os, json
from PIL import Image
from tqdm import tqdm
texts, file_names = [], []
os.makedirs("./data/lora_dataset_processed/train", exist_ok=True)
with open("./data/data-juicer/output/result.jsonl", "r") as file:
for data_id, data in enumerate(tqdm(file.readlines())):
data = json.loads(data)
text = data["text"]
texts.append(text)
image = Image.open(data["image"][0])
image_path = f"./data/lora_dataset_processed/train/{data_id}.jpg"
image.save(image_path)
file_names.append(f"{data_id}.jpg")
data_frame = pd.DataFrame()
data_frame["file_name"] = file_names
data_frame["text"] = texts
data_frame.to_csv("./data/lora_dataset_processed/train/metadata.csv", index=False, encoding="utf-8-sig")
data_frame
2.4 开始训练模型
from diffsynth import download_models
download_models(["Kolors", "SDXL-vae-fp16-fix"])
利用diffsynth库下载模型
from diffsynth import ModelManager, SDXLImagePipeline
from peft import LoraConfig, inject_adapter_in_model
import torch
def load_lora(model, lora_rank, lora_alpha, lora_path):
lora_config = LoraConfig(
r=lora_rank,
lora_alpha=lora_alpha,
init_lora_weights="gaussian",
target_modules=["to_q", "to_k", "to_v", "to_out"],
)
model = inject_adapter_in_model(lora_config, model)
state_dict = torch.load(lora_path, map_location="cpu")
model.load_state_dict(state_dict, strict=False)
return model
# Load models
model_manager = ModelManager(torch_dtype=torch.float16, device="cuda",
file_path_list=[
"models/kolors/Kolors/text_encoder",
"models/kolors/Kolors/unet/diffusion_pytorch_model.safetensors",
"models/kolors/Kolors/vae/diffusion_pytorch_model.safetensors"
])
pipe = SDXLImagePipeline.from_model_manager(model_manager)
# Load LoRA
pipe.unet = load_lora(
pipe.unet,
lora_rank=16, # This parameter should be consistent with that in your training script.
lora_alpha=2.0, # lora_alpha can control the weight of LoRA.
lora_path="models/lightning_logs/version_0/checkpoints/epoch=0-step=500.ckpt"
)
加载模型
2.5 生成图片
torch.manual_seed(0)
image = pipe(
prompt="二次元,一个灰色短发小女孩,在家中沙发上坐着,双手托着腮,很无聊,全身,蓝色连衣裙",
negative_prompt="丑陋、变形、嘈杂、模糊、低对比度",
cfg_scale=4,
num_inference_steps=50, height=1024, width=1024,
)
image.save("1.jpg")
剩下几段代码与此格式类似,均为生成图片的代码,在**pipe()**中修改参数:
- prompt:正向描述词,即你想要生成的图片应该包含的内容
- negative_prompt:反向提示词,即你不希望生成的图片的内容
- ffg_scale:控制生成图像与提示之间的匹配度,数值越高,生成的图像与提示之间的匹配度越高
- num_inference_steps:推理步骤越多,生成的图像通常会越细致和清晰
- height:生成图片的高度
- width:生成图片的宽度
最后通过**image.save()**将生成的图片保存到本地。
import numpy as np
from PIL import Image
images = [np.array(Image.open(f"{i}.jpg")) for i in range(1, 9)]
image = np.concatenate([
np.concatenate(images[0:2], axis=1),
np.concatenate(images[2:4], axis=1),
np.concatenate(images[4:6], axis=1),
np.concatenate(images[6:8], axis=1),
], axis=0)
image = Image.fromarray(image).resize((1024, 2048))
image
最后用过numpy库的**concatenate()**将图片拼接在一起,至此我们想要生成的一幅8张图片的连环画小故事就算完成了,就可以提交作品了
3 知识点补充
3.1 Data-Juicer
数据处理和转换工具,旨在简化数据的提取、转换和加载过程。该系统强大之处在于它针对多模态数据的处理,覆盖了文本、图像、音频甚至视频,为当今和未来多模态模型的发展提供了强有力的支持。
3.2 DiffSynth-Studio
高效微调训练大模型工具,它通过先进的机器学习技术,为用户提供了一种全新的创作方式,使得风格转换变得更加高效和直观。该工具的目标用户群体广泛,包括但不限于艺术家、设计师、视频编辑者和AI爱好者。