VRAM saving for image generation for FLUX

[plainline]

that fixed it for me too.

[thetashed24]

worth checking the driver first.

[quiethouse]

A couple questions before I guess too hard. What model, what runtime, and roughly what context size?

[sparrowworks88]

I tested this tonight. I couldn't reproduce the slowdown until I pushed context much higher.

[dockworks23]

I tested this tonight. I couldn't reproduce the slowdown until I pushed context much higher.

same here on my side.

[thetapath]

I hit almost the same issue last night.