I'm just testing AWS tools to serve using sagemaker. My current progress is that I've already created a trained model, and compressed pth, inference.py, requirement.txt, and other necessary files into tar.gz and deployed them to an s3 bucket.

$ tar -tzf birefnet.tar.gz
./
./ckpt/
./ckpt/BiRefNet-general-epoch_244.pth
./config.py
./dataset.py
./evaluation/
./inference.py
./requirements.txt
.
.

Here are the contents of the compressed file. After uploading to the bucket, I tried to load the tar.gz file in the bucket using pytorchmodel in the sagemaker notebook. However, I keep getting an error that inference.py is missing. I tried asking for help from chatgpt and many others, but was unsuccessful. Here is the code I wrote:

import sagemaker
from sagemaker.pytorch import PyTorchModel


sagemaker_session = sagemaker.Session()
role = sagemaker.get_execution_role()  

model_data = 's3://sagemaker-birefnet-up/path/to/birefnet.tar.gz'

pytorch_model = PyTorchModel(
    model_data=model_data,
    role=role,
    entry_point='inference.py',
    framework_version='2.5.1',
    py_version='py311',
    sagemaker_session=sagemaker_session
)

predictor = pytorch_model.deploy(
    initial_instance_count=1,
    instance_type='ml.t2.medium',
)

FileNotFoundError: [Errno 2] No such file or directory: 'inference.py'

What did I do wrong?

I thought what I wanted to do was simple. I tried recompressing and changing the model path, but nothing worked.

I'm just testing AWS tools to serve using sagemaker. My current progress is that I've already created a trained model, and compressed pth, inference.py, requirement.txt, and other necessary files into tar.gz and deployed them to an s3 bucket.

$ tar -tzf birefnet.tar.gz
./
./ckpt/
./ckpt/BiRefNet-general-epoch_244.pth
./config.py
./dataset.py
./evaluation/
./inference.py
./requirements.txt
.
.

Here are the contents of the compressed file. After uploading to the bucket, I tried to load the tar.gz file in the bucket using pytorchmodel in the sagemaker notebook. However, I keep getting an error that inference.py is missing. I tried asking for help from chatgpt and many others, but was unsuccessful. Here is the code I wrote:

import sagemaker
from sagemaker.pytorch import PyTorchModel


sagemaker_session = sagemaker.Session()
role = sagemaker.get_execution_role()  

model_data = 's3://sagemaker-birefnet-up/path/to/birefnet.tar.gz'

pytorch_model = PyTorchModel(
    model_data=model_data,
    role=role,
    entry_point='inference.py',
    framework_version='2.5.1',
    py_version='py311',
    sagemaker_session=sagemaker_session
)

predictor = pytorch_model.deploy(
    initial_instance_count=1,
    instance_type='ml.t2.medium',
)

FileNotFoundError: [Errno 2] No such file or directory: 'inference.py'

What did I do wrong?

I thought what I wanted to do was simple. I tried recompressing and changing the model path, but nothing worked.

• amazon-web-services
• pytorch
• zip
• amazon-sagemaker

Share Improve this question Follow asked Mar 25 at 8:48 dev_seodev_seo 1

Add a comment  | 

1 Answer 1

Sorted by: Reset to default 0

I think both inference.py and requirements.txt should be located in a subfolder named code, unless you specify a different folder using the parameter source_dir argument when creating PyTorch estimator.

See here for details.