PaddlePaddle / PaddleNLP

👑 Easy-to-use and powerful NLP and LLM library with 🤗 Awesome model zoo, supporting wide-range of NLP tasks from research to industrial applications, including 🗂Text Classification, 🔍 Neural Search, ❓ Question Answering, ℹ️ Information Extraction, 📄 Document Intelligence, 💌 Sentiment Analysis etc.
https://paddlenlp.readthedocs.io
Apache License 2.0
12k stars 2.92k forks source link

[Question]: It is detected that mkldnn and memory_optimize_pass are enabled at the same time, but they are not supported yet. Currently, memory_optimize_pass is explicitly disabled #5228

Closed ChinaGPT closed 6 months ago

ChinaGPT commented 1 year ago

转换模型:fd_pipe = FastDeployStableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5@fastdeploy")

出现报错:It is detected that mkldnn and memory_optimize_pass are enabled at the same time, but they are not supported yet. Currently, memory_optimize_pass is explicitly disabled

JunnYu commented 1 year ago

你好,这不是报错,建议你参考这里的部署流程,重新导出一下模型https://github.com/PaddlePaddle/PaddleNLP/tree/develop/ppdiffusers/deploy