QWEN Layerwise
This is the QWEN 3 8B LLM
Qwen layerwise is the first FP32 training with targeted layers for FLUX Klein that consistently shown improvement vs full BF16 base QWEN on Text
When scoring successes vs failure.
If both models failed on text or anatomical errors the result was neutral.
If the layerwise model improved text but caused a anatomical error this was a failure
Success was only scored if the layerwise model improved text or fixed a glaring error like number of fingers.
The success number exceeded the threshold set for what could just be viewed as random or lucky success vs base BF16.
Description
FAQ
Comments (4)
Would you ever consider doing a INT8 update to this great project. Flux Klein 9b is a great model but anatomy lets it down in my opinion other than that it's great 👍
I am not sure that that the model is the right tensor shapes for convrot
@Felldude does this text Encoder differ much from the original one or uncensored version of Qwen3 8b. As it sounds like your one can help with anatomy and stuff like that which would be very beneficial as that is one big area of weakness with it.
@AnimaXx Besides the fp32 blocks on the BF16/FP32 version it has full training yes