-
Notifications
You must be signed in to change notification settings - Fork 647
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
iOS 18 Target Regression #2353
Comments
Hey @NoahBPeterson Could you confirm on your side that |
Yes, I can confirm that it does work with
|
Here's a list of everything I have installed in pip:
|
I just created a new virtual environment and installed
|
If I set |
Here are my experiments:
It's indeed an iOS18 fp16 op regression. |
There are a lot of floating pieces. For example, I happened to find that I added this bug into our TODO list. |
Unfortunately, bisecting the model fails with both the iOS17 and iOS18 version. With this script: import coremltools as ct
model_path = "mxbai-embed-large-v1-iOS18.mlpackage"
output_dir = "./output/"
ct.models.utils.bisect_model(
model_path,
output_dir,
check_output_correctness=True
) Both the
Removing the |
🐞Describing the bug
coremltools.target.iOS17
, but only outputsnan
when targetingcoremltools.target.iOS18
To Reproduce
convert_pytorch_coreml.py
.run_coreml.py
convert_pytorch_coreml.py
tominimum_deployment_target=ct.target.iOS18,
nan
System environment (please complete the following information):
The text was updated successfully, but these errors were encountered: