Connection Error while Finetuning Phi-4 Model on Azure AI Foundry

Question

Connection Error while Finetuning Phi-4 Model on Azure AI Foundry

Nuhu Ibrahim 0

I get the connection below when fine-tuning Phi-4 model on Azure AI Foundry:

Note: I am using Managed Compute

Execution failed. User process 'Rank 0' exited with status code 1. Please check log file 'user_logs/std_log_process_0.txt' for error details. Error: [rank0]:   File "/azureml-envs/default/lib/python3.10/site-packages/requests/sessions.py", line 703, in send
[rank0]:     r = adapter.send(request, **kwargs)
[rank0]:   File "/azureml-envs/default/lib/python3.10/site-packages/requests/adapters.py", line 700, in send
[rank0]:     raise ConnectionError(e, request=request)
[rank0]: exceptions.PredictException: PredictException:
[rank0]: 	Message: Model Prediction failed due to [ConnectionError(MaxRetryError("HTTPConnectionPool(host='0.0.0.0', port=8000): Max retries exceeded with url: /generate (Caused by NewConnectionError('

Saideep Anchuri 7,040 Reputation points Microsoft External Staff Moderator

2025-04-23T19:15:54.9766667+00:00
Hi Nuhu Ibrahim

The connection error you are encountering while fine-tuning the Phi-4 model on Azure AI Foundry.

Here are some steps:

Ensure that your network settings allow for the necessary connections. If you are using a managed virtual network, check that the settings are correctly configured to allow access to the required resources.

Verify that the Azure Machine Learning compute cluster you selected for fine-tuning is running and has sufficient resources available.

Review the log file mentioned in the error message (user_logs/std_log_process_0.txt) for more specific details about the failure.

Retry the Process: If the error persists, you may consider retrying the fine-tuning process to see if it was a temporary issue.

Kindly refer below link: using-managed-compute

Thank You.
Prashanth Veeragoni 4,440 Reputation points Microsoft External Staff Moderator

2025-04-24T18:41:26.7066667+00:00

Hi Nuhu Ibrahim,

Following up to see if the above suggestion was helpful. And, if you have any further query do let us know.

Thank you!
Saideep Anchuri 7,040 Reputation points Microsoft External Staff Moderator

2025-04-25T16:31:12.84+00:00

Hi Nuhu Ibrahim,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet.

Thank You.
Saideep Anchuri 7,040 Reputation points Microsoft External Staff Moderator

2025-04-30T13:25:41.59+00:00

Hi Nuhu Ibrahim,

Did you get any chance to check private message.

Thank you.

1 answer

Your answer

Saideep Anchuri 7,040 Reputation points Microsoft External Staff Moderator

2025-04-23T19:15:54.9766667+00:00

Hi Nuhu Ibrahim

The connection error you are encountering while fine-tuning the Phi-4 model on Azure AI Foundry.

Here are some steps:

Ensure that your network settings allow for the necessary connections. If you are using a managed virtual network, check that the settings are correctly configured to allow access to the required resources.

Verify that the Azure Machine Learning compute cluster you selected for fine-tuning is running and has sufficient resources available.

Review the log file mentioned in the error message (user_logs/std_log_process_0.txt) for more specific details about the failure.

Retry the Process: If the error persists, you may consider retrying the fine-tuning process to see if it was a temporary issue.

Kindly refer below link: using-managed-compute

Thank You.
Prashanth Veeragoni 4,440 Reputation points Microsoft External Staff Moderator

2025-04-24T18:41:26.7066667+00:00

Hi Nuhu Ibrahim,

Following up to see if the above suggestion was helpful. And, if you have any further query do let us know.

Thank you!
Saideep Anchuri 7,040 Reputation points Microsoft External Staff Moderator

2025-04-25T16:31:12.84+00:00

Hi Nuhu Ibrahim,

We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet.

Thank You.
Saideep Anchuri 7,040 Reputation points Microsoft External Staff Moderator

2025-04-30T13:25:41.59+00:00

Hi Nuhu Ibrahim,

Did you get any chance to check private message.

Thank you.

Answer 1

Hi all, thanks so much for your response. I tried to finetune the model again on managed compute but it shows same error. Also, I couldn't do any error tracing further because it is managed compute - which means I don't have access to the infrastructure. I was only able to fill in the parameters through the interface and the entire training process is handled my azure ai foundry in the background.

As I couldn't get it fixed, I resolved to other serverless models. However, it would be nice to know how to solve issues like these in the future considering cases may come when particular models are only available on managed compute and not on serverless.

Thanks,
Nuhu

Share via

Connection Error while Finetuning Phi-4 Model on Azure AI Foundry

1 answer

Your answer