CODEWITHSUNDEEP
amazon-web-servicestensorflowmachine-learningamazon-sagemaker
Malachi Erskine
Malachi Erskine

Reputation: 1

deploying a Tensorflow model artifact to Sagemaker

I am trying to deploy a TensorFlow model to a Sagemaker endpoint. I have the model artifact at generic_graph.pb and the model's labels at labels.txt.

I started by creating a tar file with the following contents:

# #model directory structure 
# #model.tar.gz
# └── <model_name>
#     └── <version_number>
#         ├── saved_model.pb
#         └── variables
#             ├── labels.txt

I uploaded the file to a bucket in S3. Then, I tried to deploy the model with the following code:

sagemaker_session = sagemaker.Session()
role = 'my-role'

model = TensorFlowModel(model_data='s3://my-bucket/model.tar.gz',
                    role=role,
                    framework_version='2.3.0')


predictor = model.deploy(initial_instance_count=1, instance_type='ml.m5.large')

I keep getting the following error in my cloudwatch logs:

ValueError: no SavedModel bundles found!

Not sure what else to try.

Upvotes: 0

Views: 30

Answers (0)

Related Questions