Unable to startup the server for image classification

[I-aashishM]

When I run the script below, the error occur

deepsparse.server \
    --model_path "zoo:cv/classification/resnet_v1-50/pytorch/sparseml/imagenet/pruned-moderate" \
    --task image_classification 

ValueError: unsupported task given of image_classification for serve model config task='image_classification' model_path='zoo:cv/classification/resnet_v1-50/pytorch/sparseml/imagenet/pruned-moderate' batch_size=1 alias=None kwargs={} engine='deepsparse' num_cores=None scheduler='async' .

[rahul-tuli]

Hi @I-aashishM, which version of deepsparse are you working with? Can you paste the output from pip show deepsparse, along with your OS and the installation pathway you used to install deepsparse?

[I-aashishM]

pip3 show deepsparse

Name: deepsparse Version: 0.12.2 Summary: Neural network inference engine that delivers GPU-class performance for sparsified models on CPUs Home-page: https://github.com/neuralmagic/deepsparse Author: Neuralmagic, Inc. Author-email: [email protected] License: Neural Magic Engine License, Apache Location: {$HOME}/cpu_perf/lib/python3.6/site-packages Requires: numpy, onnx, protobuf, requests, sparsezoo, tqdm

OS : ubuntu 18.04

I installed the deepsparse using command below: pip3 install deepsparse

and, It is install in {$HOME}/cpu_perf/lib/python3.6/site-packages (virtual environment)

[I-aashishM]

I have updated the latest deepsparse version. The output of pip3 show deepsparse is below:

Name: deepsparse Version: 1.0.1 Summary: Neural network inference engine that delivers GPU-class performance for sparsified models on CPUs Home-page: https://github.com/neuralmagic/deepsparse Author: Neuralmagic, Inc. Author-email: [email protected] License: Neural Magic Engine License, Apache Location: {$HOME}/CPu_performance/cpu_perf/lib/python3.6/site-packages Requires: click, numpy, onnx, protobuf, pydantic, requests, sparsezoo, tqdm

After I run the below command : deepsparse.server \ --model_path "zoo:cv/classification/resnet_v1-50/pytorch/sparseml/imagenet/pruned-moderate" \ --task image_classification

Now I'm getting a different error:

INFO: Started server process [6096] INFO: Waiting for application startup. INFO: Application startup complete. INFO: Uvicorn running on http://0.0.0.0:5543 (Press CTRL+C to quit) INFO: 127.0.0.1:55920 - "GET / HTTP/1.1" 307 Temporary Redirect INFO: 127.0.0.1:55920 - "GET /docs HTTP/1.1" 200 OK INFO: 127.0.0.1:55920 - "GET /openapi.json HTTP/1.1" 500 Internal Server Error ERROR: Exception in ASGI application

ValueError: Value not declarable with JSON Schema, field: name='images_ndarray' type=ndarray required=True

[rahul-tuli]

Hi @I-aashishM, We have dropped python3.6 support kindly use a version b/w 3.7 - 3.10

Kindly follow the following steps to run the server (I verified these on my end) Assuming inside an activated virtual environment

• Install server dependencies

pip install deepsparse[server]

• Run the server

deepsparse.server \ 
    --model_path "zoo:cv/classification/resnet_v1-50/pytorch/sparseml/imagenet/pruned-moderate" \
    --task image_classification

This should start a working server at http://0.0.0.0:5543 by default

Closing this issue for now, kindly let us know if you see this issue again. Thanks!