-
Notifications
You must be signed in to change notification settings - Fork 117
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
iam unable to load model #142
Comments
3060 may require sequence CPU offloading, 55GB should be sufficient |
the same goes for 3090 ... strange... the predict_i2v.py is extremely slow when loading the model into gpu and when I run 'watch nvidia-smi' i cannot even see the model loading inside memory, I haven't experience something like this before... Is this normal behavior? the files nor the model aren't that big... |
Did you use qfloat8 before? |
Yes. apart from Everything it even not loading model right
Yes. apart from Everything it even not loading model right |
No, but I was running the code with GPU_memory_mode = "model_cpu_offload" not with model_cpu_offload_and_qfloat8 - I don't know whether you are using qfloat8 in model_cpu_offload, however gonna try model_cpu_offload_and_qfloat8 |
Sorry, the 3090 needs to use model_cpu_offload_and_qfloat8 due to insufficient VRAM. |
We are currently trying to develop a smaller model to meet the low memory limit (30GB) |
30GB RAM will be support in #154 |
i have 55gb ram and 3060gpu
The text was updated successfully, but these errors were encountered: