As someone informed me, the turbo LoRa is a bit huge. It's 600mb and some of those layers don't do anything but affect flux guidance and make the outputs shiny and plastic. So why not go with what I've learned and just excise anything that is related to double_blocks.
The result is now 300mb and I seem to get less plastic/body horrors. I'm genning on 1.9 flux CFG, worked on openflux to reduce the steps some. Can't get anything at 4 steps but the original doesn't either.
描述:
This is the BF16 version. All double_block layers were cut out. I made another one with only single blocks but it seems the same and saved 30mb so no point. Maybe only lora_a portion can be used or something.. might be worth trying.
训练词语:
名称: flux-turbo-small-bf16.sft
大小 (KB): 312944
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
