The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.
Commits on master
Support loading and using SkyReels-V1-Hunyuan-I2V (#6862)
acc152b Jukka Seppänen committed 1 year ago
Add a node to set the model compute dtype for debugging.
2e21122 comfyanonymous committed 1 year ago
Auto enable mem efficient attention on gfx1100 on pytorch nightly 2.7
d7b4bf2 comfyanonymous committed 1 year ago
Open yaml files with utf-8 encoding for extra_model_paths.yaml (#6807)
042a905 Robin Huang committed 1 year ago
Add a way to set a different compute dtype for the model at runtime.
019c702 comfyanonymous committed 1 year ago
Better memory estimation for ROCm that support mem efficient attention.
8773ccf comfyanonymous committed 1 year ago
Make --force-fp16 actually force the diffusion model to be fp16.
af4b7c9 comfyanonymous committed 1 year ago