You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The Git homepage mentions that the project can be run without using Apex.
#set enable_layernorm_kernel=False in config to avoid using Apex
What is the specific command for this?
I added --enable_layernorm_kernel=False after the inference command, but it reported an error that the parameter was unrecognized.
I also modified the 'enable_layernorm_kernel': False under the script opensora/configs/opensora/inference/16x256x256.py, but it still doesn't work.
The text was updated successfully, but these errors were encountered:
Hi, you need to navigate to the file "configs/opensora-v1-1/train/stage2.py" and make the following modifications:
model = dict(
type="STDiT2-XL/2",
from_pretrained=None,
input_sq_size=512, # Note: the pretrained model is trained on 512x512
qk_norm=True,
enable_flashattn=True,
enable_layernorm_kernel=False,
)
The Git homepage mentions that the project can be run without using Apex.
#set enable_layernorm_kernel=False in config to avoid using Apex
What is the specific command for this?
I added --enable_layernorm_kernel=False after the inference command, but it reported an error that the parameter was unrecognized.
I also modified the 'enable_layernorm_kernel': False under the script opensora/configs/opensora/inference/16x256x256.py, but it still doesn't work.
The text was updated successfully, but these errors were encountered: