ComfyUi crashes when running big loras on Flux

[TermTo]

Expected Behavior

I have no issues loading and running loras when they aren't that big.

Actual Behavior

When the lora is too big (around 1GB), ComfyUi simply crashes, for example, I can't run this lora because of that: https://huggingface.co/ByteDance/Hyper-SD/blob/main/Hyper-FLUX.1-dev-16steps-lora.safetensors

Steps to Reproduce

Here's my workflow:

Debug Logs

That's the problem, there are no message errors when it crashes, I simply have this:

got prompt
  6%|█████▏                                                                             | 1/16 [00:05<01:23,  5.60s/it]
D:\ComfyUI_windows_portable>pause
Press any key to continue . . .

Other

No response

[JorgeR81]

Can you use other big loras, like this one ( 1.28 GB )  https://civitai.com/models/641309/formcorrector-anatomic?modelVersionId=717317

I can use it with Flux FP8 and  --reserve-vram 1.2 Or with GGUF models, without any flags. https://github.com/city96/ComfyUI-GGUF

[JorgeR81]

Just tested:

• GTX 1070 ( 8GB )
• 32 GB VRAM
• Windows 10

Hyper-FLUX.1-dev-8steps-lora.safetensors works for me in Flux FP8 and --reserve-vram 1.2

[BrechtCorbeel]

I don't know what is happening today, but lately flux goes haywire on me all the time, it used to work fine, but I can barely get 4 images out every two hours with mountains of errors and issues that did not used to be there all related to memory.

[JorgeR81]

Yeah, I think there is an issue on the GGUF side. https://github.com/city96/ComfyUI-GGUF/issues/75

---

EDIT: this issue, with GGUF models, seems to be fixed now.