Example: best combination for Diffusers
Currently, one of the base Smash_Config for the Diffusers model is:
Disclaimer: the "best" config always depends on your use case.
smash_config["distributer"] = "ring_attn" smash_config["cacher"] = "auto" smash_config["compiler"] = "torch_compile" smash_config["factorizer"] = "qkv_diffusers" smash_config["quantizer"] = "fp8" smash_config["pruner"] = "padding_pruning"
This is currently used on Flux Dev on Replicate: https://replicate.com/prunaai/flux.1-dev
Do not hesitate to contact us (support@pruna.ai) for assistance deploying this configuration in your environment.