PaddlePaddle / PaddleMIX

Paddle Multimodal Integration and eXploration, supporting mainstream multi-modal tasks, including end-to-end large-scale multi-modal pretrain models and diffusion model toolbox. Equipped with high performance and flexibility.
Apache License 2.0
357 stars 140 forks source link

【PaddleMIX 热身打卡】跑通 Stable-Diffusion 训练报错 #806

Closed fxy1699 closed 1 week ago

fxy1699 commented 1 week ago

273

ppdiffusers-0.29.0,安装方式为python setup.py install 在按照文档对laion400数据集进行训练时运行train.sh脚本后报错ValueError: Not supported value: 0,查看对应源码/home/aistudio/.local/lib/python3.10/site-packages/ppdiffusers-0.29.0-py3.10.egg/ppdiffusers/utils/constants.py后,尝试修改train.sh环境变量值0为false,仍然报错。 train.sh如下 image 报错如下: image /opt/conda/envs/python35-paddle120-env/lib/python3.10/site-packages/_distutils_hack/__init__.py:26: UserWarning: Setuptools is replacing distutils. warnings.warn("Setuptools is replacing distutils.") Traceback (most recent call last): File "/home/aistudio/PaddleMIX/ppdiffusers/examples/stable_diffusion/train_txt2img_laion400m_trainer.py", line 21, in <module> from sd import ( File "/home/aistudio/PaddleMIX/ppdiffusers/examples/stable_diffusion/sd/__init__.py", line 17, in <module> from .sd_args import SDDataArguments, SDModelArguments, SDTrainingArguments File "/home/aistudio/PaddleMIX/ppdiffusers/examples/stable_diffusion/sd/sd_args.py", line 31, in <module> if str2bool(os.getenv("FLAG_FUSED_LINEAR", "False")): File "/home/aistudio/.local/lib/python3.10/site-packages/ppdiffusers-0.29.0-py3.10.egg/ppdiffusers/utils/constants.py", line 30, in str2bool raise ValueError("Not supported value: {}".format(v)) ValueError: Not supported value: false

fxy1699 commented 1 week ago

pip install ppdiffusers也尝试过,是相同的错误

luyao-cv commented 1 week ago

https://github.com/PaddlePaddle/PaddleMIX/blob/develop/ppdiffusers/requirements.txt 请使用ppdiffusion 0.23.0版本哈~0.29的太高啦

fxy1699 commented 1 week ago

https://github.com/PaddlePaddle/PaddleMIX/blob/develop/ppdiffusers/requirements.txt 请使用ppdiffusion 0.23.0版本哈~0.29的太高啦

没找到0.23.0版本的,切换到了0.24.0上述问题已经解决 git checkout ppdiffusers0.24.0 image 在该branch下重新安装环境(aistudio停止后重新启动),在ppdiffusers目录下使用 pip install -e .安装环境,运行train.sh,报错如下: image huggingface-hub版本为0.23.4

luyao-cv commented 1 week ago

paddlenlp==v3.0.0-beta0 huggingface_hub==0.23.0 huggingface_hub降下版本,试下0.23和0.22哈。