Stuck at loading 1 model, and RAM dedicated stuck at max.

[naK-GH]

---

Issue/Feature Request Description

Type of Issue:

• Bug Stuck loading 1 model and RAM usage

Steps to Reproduce (For Bugs Only)

1. Windows setup.bat Python 3.10 ; Nvidia 4070Ti
2. Launch a generation and it stuck. Sometimes i have the result sometime no, but i'm always stuck on cmd at Loading 1 model, and gpu ram usage stuck at 12Go ( 4070Ti )

---

Environment

• OS: Win 11
• Version: 24H2
• Any Relevant Dependencies: 4070Ti. 32Go RAM; 13700K hyperthreading off

---

[naK-GH]

On screenshot when it's stuck at model 1, i tried to generte an other image. That's why you can see the INFO line.

[ForteDexe]

Same issue, but in Linux. What is up with this:

We will use 90% of the memory on device 0 for storing the model, and 10% for the buffer to avoid OOM. You can set `max_memory` in to a higher value to use more memory (at your own risk).

I don't know where it comes from, but certainly it prevent me generate because of run out of memory

[zliucz]

Check this solution to oom. You may disable the llava module.