4 Commits

Author	SHA1	Message	Date
leejet	f5997a1951	fix: do not force using f32 for some flux layers ... This sometimes leads to worse result	2024-08-25 14:07:22 +08:00
leejet	1bdc767aaf	feat: force using f32 for some layers	2024-08-25 13:53:16 +08:00
leejet	c837c5d9cc	style: format code	2024-08-25 00:19:37 +08:00
leejet	64d231f384	feat: add flux support (#356) ... * add flux support * avoid build failures in non-CUDA environments * fix schnell support * add k quants support * add support for applying lora to quantized tensors * add inplace conversion support for f8_e4m3 (#359) in the same way it is done for bf16 like how bf16 converts losslessly to fp32, f8_e4m3 converts losslessly to fp16 * add xlabs flux comfy converted lora support * update docs --------- Co-authored-by: Erik Scholz <Green-Sky@users.noreply.github.com>	2024-08-24 14:29:52 +08:00