Parameter validation failed: Unknown parameter in PrimaryContainer

[mak-rayate]

Describe the bug
I'm trying to deploy meta-textgeneration-llama-codellama-7b jumpstart model endpoints using python. In sage-maker notebook code works fine. But deployed same code in app.py file giving an error of ParamValidationError.
Tried --> installed upgraded boto3 and botocore lib.

System
Python : 3.9.15
SageMaker Version: 2.214.2
boto3 : Version: 1.34.76
botocore : Version: 1.34.76
streamlit :Version: 1.32.2

To reproduce
app.py file code :

import streamlit as st

def main():
st.title('LLAMA Speak')

  #Generate code LLAMA 2 model endpoints
  sideb = st.sidebar
  check1 = sideb.button("generate endpoint")
  check2 = sideb.button("kill")
  if check1:
      endpoint_name = "code-llama-7b-11"
      from sagemaker.jumpstart.model import JumpStartModel
      model1 = JumpStartModel(model_id="meta-textgeneration-llama-codellama-7b")
      predictor = model1.deploy(
          endpoint_name=endpoint_name,
          accept_eula = True
          # instance_type="ml.g5.12xlarge", # uncomment to use another instance
          )
      st.info("Code LLAMA2 endpoints has been generated successfully !")
  if check2:
      predictor.delete_model()
      predictor.delete_endpoint()
      st.info("Code LLAMA2 endpoints stopped successfully ! ")

if name == 'main':
main()

Logs
ParamValidationError: Parameter validation failed: Unknown parameter in PrimaryContainer: "ModelDataSource", must be one of: ContainerHostname, Image, ImageConfig, Mode, ModelDataUrl, Environment, ModelPackageName, InferenceSpecificationName, MultiModelConfig
Traceback:
File "/opt/conda/lib/python3.9/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 542, in _run_script
exec(code, module.dict)
File "/home/sagemaker-user/natural_to_SQL/app.py", line 36, in
main()
File "/home/sagemaker-user/natural_to_SQL/app.py", line 21, in main
predictor = model1.deploy(
File "/opt/conda/lib/python3.9/site-packages/sagemaker/jumpstart/model.py", line 638, in deploy
predictor = super(JumpStartModel, self).deploy(**deploy_kwargs.to_kwargs_dict())
File "/opt/conda/lib/python3.9/site-packages/sagemaker/model.py", line 1610, in deploy
self._create_sagemaker_model(
File "/opt/conda/lib/python3.9/site-packages/sagemaker/jumpstart/model.py", line 467, in _create_sagemaker_model
super(JumpStartModel, self)._create_sagemaker_model(
File "/opt/conda/lib/python3.9/site-packages/sagemaker/model.py", line 914, in _create_sagemaker_model
self.sagemaker_session.create_model(**create_model_args)
File "/opt/conda/lib/python3.9/site-packages/sagemaker/session.py", line 3887, in create_model
self._intercept_create_request(create_model_request, submit, self.create_model.name)
File "/opt/conda/lib/python3.9/site-packages/sagemaker/session.py", line 6427, in _intercept_create_request
return create(request)
File "/opt/conda/lib/python3.9/site-packages/sagemaker/session.py", line 3875, in submit
self.sagemaker_client.create_model(**request)
File "/opt/conda/lib/python3.9/site-packages/botocore/client.py", line 565, in _api_call
return self._make_api_call(operation_name, kwargs)
File "/opt/conda/lib/python3.9/site-packages/botocore/client.py", line 974, in _make_api_call
request_dict = self._convert_to_request_dict(
File "/opt/conda/lib/python3.9/site-packages/botocore/client.py", line 1048, in _convert_to_request_dict
request_dict = self._serializer.serialize_to_request(
File "/opt/conda/lib/python3.9/site-packages/botocore/validate.py", line 381, in serialize_to_request
raise ParamValidationError(report=report.generate_report())

[ahmedselhady]

@mak-rayate have you been able to solve this issue? I am facing the same problem.

[mak-rayate]

Hello @ahmedselhady ,
No the error is still open but I have found alternative way (not recommended) to call the model. Instead of deploy model using python, deploy endpoint inside Jumpstart and use it in your code as below :
llm_codellama = SagemakerEndpoint(
endpoint_name='jumpstart-dft-meta-textgeneration-llama-codellama-7b', <-- paste model endpoint here generated by Jumpstart interface
region_name=region,
model_kwargs=parameters,
endpoint_kwargs={"CustomAttributes": 'accept_eula=true'},
content_handler=content_handler
)