Replies: 6 comments
-
@ashkanaev : You could post this question on discuss.mxnet.io to get a wider audience. Also, are you seeing errors when you modify the predictor or is it not adhering to your modified input shapes? It would be helpful to know what you mean by making it flexible. @sandeep-krishnamurthy : Could you tag this with "C++", "Example" and "Question". |
Beta Was this translation helpful? Give feedback.
-
MXPredReshape |
Beta Was this translation helpful? Give feedback.
-
I am calling MXPredReshape with an updated input_shape_data (only changing the batch size), but when I do, the MXPredSetInput tells me my data size is larger than the expected shape. I am using the new handle, and even freeing the old handle. Can someone check to see if this is a bug in c_predict_api.cc or if I'm somehow messing it up. |
Beta Was this translation helpful? Give feedback.
-
@ashkanaev thank you for submitting this issue! @nswamy can you please update the labels? |
Beta Was this translation helpful? Give feedback.
-
@leleamol could you pick this up ? Thanks! |
Beta Was this translation helpful? Give feedback.
-
@ashkanaev Is this problem solved? How to achieve it? |
Beta Was this translation helpful? Give feedback.
-
Hello!
Is't possible to make predictor with c++ API for arbitrary batch size?
From https://github.com/apache/incubator-mxnet/blob/97570916f844bcb4515d972c75fb0a75da345d97/example/image-classification/predict-cpp/image-classification-predict.cc
const mx_uint input_shape_data[4] = { 1, static_cast<mx_uint>(channels), static_cast<mx_uint>(height), static_cast<mx_uint>(width) };
MXPredCreate(static_cast<const char*>(json_data.GetBuffer()), static_cast<const char*>(param_data.GetBuffer()), static_cast<int>(param_data.GetLength()), dev_type, dev_id, num_input_nodes, input_keys, input_shape_indptr, input_shape_data, &pred_hnd);
look's like predictor hard linked to fixed input data shape. how to make it flexible?
Beta Was this translation helpful? Give feedback.
All reactions