6 Commits

Author SHA1 Message Date
comfyanonymous
19c503d13f Lower fp8 lora memory usage. 2024-09-03 01:25:05 -04:00
comfyanonymous
f2a13e96b4 Make the stochastic fp8 rounding reproducible. 2024-08-26 15:12:06 -04:00
comfyanonymous
6c7971a4d7 Use less memory in float8 lora patching by doing calculations in fp16. 2024-08-26 14:45:58 -04:00
comfyanonymous
f99d236347 Better subnormal fp8 stochastic rounding. Thanks Ashen. 2024-08-19 13:38:03 -04:00
comfyanonymous
7921766c26 Handle subnormal numbers in float8 rounding. 2024-08-19 05:51:08 -04:00
comfyanonymous
1fb6ab1bcd Fix loras having a weak effect when applied on fp8. 2024-08-17 15:20:17 -04:00