Daniil Goncharov comments

Results 123 comments of


                                            Daniil Goncharov

Running target operation for training in C++

Unfortunately, I did not work with ungrozen graph. I usually use python to train, and use on c ++ then to run.

Running target operation for training in C++

@yangjituan you can also check https://gist.github.com/asimshankar/7c9f8a9b04323e93bb217109da8c7ad2.

read checkpoint instead of pb file

I found how read checkpoint in C API. Will add example soon.

read checkpoint instead of pb file

Hi, I add read checkpoint. Could you please test new function `TF_Graph* LoadGraph(const char* graph_path, const char* checkpoint_prefix)`?

Memory leak during inference with frozen graph

Hi @Romit-Maulik, could you please check https://github.com/Neargye/hello_tf_c_api/commit/04fb01fca87adeb09137e0a32445d4374a702ca0?

Memory leak during inference with frozen graph

I will check later.

Memory leak during inference with frozen graph

Hi @apivovarov, could you give an example of your code? To reduce the number of allocations, you can use the following example ```cpp auto sess = TF_NewSession(graph, options, status); auto...

Memory leak during inference with frozen graph

@apivovarov could you try change line 42 to ``` TF_Tensor* output_tensors[1] = {TF_AllocateTensor(TF_FLOAT, output_dims, output_dims_size, output_num_bytes)}; ```

Create Tensor of strings (vector<char*>) with C API

I will fix api soon, to support string tensor. Could you help me test this?

image prediction

You need to load the image as an array of float, load this array into the input tensor, and run session. Later I will add an example of a images...