Hi @<1715900760333488128:profile|ScaryShrimp33>
hi everyone! I’m trying to save my model’s weights to storage. And I can’t do it.
See example here: None
or
task.update_output_model(model_path="/path/to/model.pt")
hi everyone! I’m trying to save my model’s weights to storage. And I can’t do it.
I’m tried upload_artifact(), both, with python objects or path to file. There are errors in uploading and downloading files.
Usually with file size. But it’s a small resnet18 models, ~40mb
I’m trying to upload them like this:
”model_test = models.resnet18(weights='IMAGENET1K_V1')
task_name = "test_uploading"
task = Task.init(project_name=f'ai_watches/{task_name}', task_name=f'test_3')
task.upload_artifact('model_fromobject', model_test.state_dict(), wait_on_upload=False)
”
or
”model_path = "models/test_uploading/model_test_nodropout.pt"
model_test = models.resnet18(weights='IMAGENET1K_V1')
torch.save(model_test.state_dict(), model_path)
task.upload_artifact('model_fromofile', model_path, wait_on_upload=False)
”
Pro subscription, default clean-ml services and files. Storage has a lot of space. I easily upload there datasets or DFs as artifacts, but can’t upload even a small model.
What I’m doing wrong?
Hi @<1715900760333488128:profile|ScaryShrimp33>
hi everyone! I’m trying to save my model’s weights to storage. And I can’t do it.
See example here: None
or
task.update_output_model(model_path="/path/to/model.pt")