Thomas Delteil

Dropping a few things that I would love to have in Gluon CV, though I realize some might not be realistic for 0.5. They don't have issue associated with them...

Thanks for the great contributions guys :+1:

This solved my issue and limited the number of workers starting to what my GPU could handle.

The idea is to let the consumer of the API know what version of the model it is consuming, to help with debugging for example. Being able to hit the...

This a Jetson TX2 which is an edge device, it isn't running on AWS. One work around is to create a executable `nvidia-smi` file in `usr/local/bin` that contains the following:...

I can't get a JSON request to go through: ``` [INFO 2018-05-10 00:23:25,486 PID:87660 /home/ubuntu/anaconda3/lib/python3.6/site-packages/mms/serving_frontend.py:predict_callback:423] Request input: data should be json tensor. [INFO 2018-05-10 00:23:25,486 PID:87660 /home/ubuntu/anaconda3/lib/python3.6/site-packages/mms/request_handler/flask_handler.py:get_form_data:111] Getting form data...

Ok so there is a lack of clarity here IMO in the `signature.json` file. When you specify input type: application/json, it seems to mean that you will actually need send...

@lupesko yes I have figured that out now. To rephrase my issue (and current understanding) is: - The `signature.json` file lets you define an `input_type` as `application/json`. - However MMS...

@vdantu thanks I see, however I think that MMS being what it is now and where it is going, isn't so tied to python anymore. Maybe worth re-considering in the...

I think I have the same problem, can't run the GPU tests after building with CUDA 9.2 ``` ./test_gpu Running GPU tests terminate called after throwing an instance of 'std::runtime_error'...