Error Message/Code: Expected argument current_run to have a valid value.

[mastreips]

I am trying to modify the following notebook tutorial in the AML environment - /classification-text-dnn/auto-ml-classification-text-dnn.ipynb so that it outputs and saves the model in ONNX format. I was able to save to pickle format, but when I try to save the output to ONNX using:

automl_run.get_output(return_onnx_model=True)

I get the following error:

ValidationException:
Message: Encountered an internal AutoML error. Error Message/Code: Expected argument current_run to have a valid value.

I know that only certain models work with Onnx conversion through AutoML (reference: https://docs.microsoft.com/en-us/azure/machine-learning/how-to-configure-auto-train#configure-your-experiment-settings) I attempted to prevent any compatibility issues via my AutoML settings below:

here is my AutoML config:

`automl_settings = {
"experiment_timeout_minutes": 20,
"experiment_exit_score": 0.70,
"primary_metric": 'accuracy',
"max_concurrent_iterations": num_nodes,
"max_cores_per_iteration": -1,
"enable_dnn": False,
"enable_early_stopping": True,
"validation_size": 0.3,
"verbosity": logging.INFO,
"enable_voting_ensemble": False,
"enable_stack_ensemble": False,
}

automl_config = AutoMLConfig(task = 'classification',
debug_log = 'automl_errors.log',
compute_target=compute_target,
training_data=train_dataset,
label_column_name=target_column_name,
blocked_models = ['LightGBM', 'Averaged Perceptron Classifier'],
enable_onnx_compatible_models=True,
**automl_settings
)`

[ghost]

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @azureml-github.

[rtanase]

@mastreips, I believe no ONNX models were generated for the experiment. To be 100% sure, do you mind sharing the run id for which you tried to download the ONNX model?

[YunsongB]

@mastreips, we have identified this issue a couple of days before and released the fix in the azureml-automl-runtime version 1.16.0. Could you please upgrade the automl package(s) and retry?

[cartacioS]

Hi @mastreips can you please let us know if the fix above worked or if you are still facing the issue?

[cartacioS]

#please-close