AI practical training camp&MMagic installation and configuration
Introduction to MMagic
MMagic is an open source toolbox for image & video editing and generation based on PyTorch. Is a member of the OpenMMLab project.
Currently, MMagic supports a variety of image and video generation/editing tasks.
Main features
- SOTA Algorithm
MMagic provides SOTA algorithms for processing, editing, and generating images and videos.
The powerful and popular application
MMagic supports popular tasks such as image restoration, image and text generation, 3D generation, image repair, cutout, super-resolution and generation. In particular, MMagic supports fine-tuning of Stable Diffusion and many exciting diffusion applications, such as ControlNet animation generation. MMagic also supports interpolation, projection, editing and other popular applications of GANs. Please start your AIGC exploration journey now! - Efficient framework
Through the MMEngine and MMCV of the OpenMMLab 2.0 framework, MMagic decomposes the editing framework into different components, and custom editor models can be easily built by combining different modules. We can define the training process like building "Lego" and provide a wealth of components and strategies. In MMagic, you can use different APIs to fully control the training process. Thanks to MMSeparateDistributedDataParallel, distributed training of dynamic model structures can be easily implemented.
2. Install MMagic
-
First, you must check your pytorch version and python version, because this library is built based on python3.8+ and pytorch1.10+**. If it is lower than this version, various inexplicable errors will occur.
-
Next, we install the MMCV and MMEngine environments, because we have installed these two libraries in the previous environment. Let’s briefly talk about it here.
-
Next we use mim to install mmagic
-
We have to download mmagic locally
-
After entering this environment, compile it
-
If the following error occurs, the solution is as follows:
-
Solution:
Installation completed! ! !