axel578

Results 33 comments of axel578

Broken training that appeared suddenly

torch-2.1.2+cu118 i my version installed, I'll reinstall it

NaN losses after hours of training (UPSAMPLER)

Same issue for me

NaN losses after hours of training (UPSAMPLER)

@lucidrains any reasons it could lead to that issue ?

problems with many python example dependencies

I have the exact same issue as listed above

System.ArgumentException: Value does not fall within the expected range on NVIDIA RTX A2000

Same here

Flux Lora from SimpleTuner training not working with ComfyUI

same issue

Very High VRAM usage when using lora with flux

> Try updating: update/update_comfyui.bat if you are on the standalone. I updated it just now an it still uses 10 Gb for a a single lora rank 64: ![image](https://github.com/user-attachments/assets/cd7a92b7-a037-4119-8fa4-92dc44962375) (Yes...

Very High VRAM usage when using lora with flux

Its still problematic on last version @comfyanonymous

Slowness with lora and control net (any control net model) for Flux

> Update your pytorch to at least 2.3 and your nvidia drivers to the latest. I update to 2.3.1 and latest driver and the exact same issue occur with very...

Slowness with lora and control net (any control net model) for Flux

> Can you check if things have improved on the latest commit? ![image](https://github.com/user-attachments/assets/59fc7c9e-f58f-4dfe-929f-7d0201f348fd) Its the exact same issue, on latest commit you did for fp8 lora, I used your 2.0...