[Bug]: No model is selected. Skip packaging output artifacts. whisper about olive HOT 8 CLOSED

re installing everything solved the issue. Thanks for the support

from olive.

would you please help double check the available EPs? from the logging, it seems the cuda EP is not available.

from olive.

hello how can I check that knowing that the following is the result of nvidia-smi ?

Mon Sep 18 05:50:27 2023
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.86.05 Driver Version: 535.86.05 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 Tesla V100S-PCIE-32GB Off | 00000000:00:06.0 Off | 0 |
| N/A 31C P0 26W / 250W | 0MiB / 32768MiB | 0% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+

+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| No running processes found |

from olive.

Just run onnxruntime.get_available_providers would return the available providers.

from olive.

how strange....

onnxruntime.get_available_providers()
['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'AzureExecutionProvider', 'CPUExecutionProvider']

from olive.

and the result of pip freeze | grep onnx && pip freeze | grep ort is
onnx==1.14.0
onnxruntime-extensions==0.8.0
importlib-resources==6.0.1
ort-nightly-gpu==1.16.0.dev20230903001

from olive.

@jsoto-gladia, would you please use a local onnx model to try the CUDA EP like:

session = onnxruntime.InferenceSession(
        model, providers=['CUDAExecutionProvider', 'CPUExecutionProvider']
)
print(session.get_providers())

From the current log, it seems the output only contains CPU EP. If this is the case, some of the cuda EP setting is not correct configured.

from olive.

Follow above setup procedure, whisper case worked well for me.

Just ensure the code snippet that Mike provide can work with CUDA EP.

from olive.