vivianoriccardo / learning-lab-c-library Goto Github PK

This library is by far the most comprehensive deep learning framework in C, and is strangely underrated.

One possible reason is that the examples have not been updated for the new versions, and thus, newcomers cannot appreciate the performance of this library.

function itoa has been changed to itoa_n.

function fully_connected needs mode as the 11th argument.

I was trying to run some of the tests like test1 through test6 and I'm noticing that fully_connected needs two extra parameters.
So I tried putting in for example 8, NO_NORMALIZATION as the last two parameters, however I don't know if that is supposed to work.
I'm able to compile however there is an Illegal instruction (core dumped) on my Ubuntu platform.
Specifically with a debugger it is showing that c->layer = layer in convolutional_layers.c is a problem.

I found out that during model training RAM usage increase continuously.
Deep dive into sources show that in function model_tensor_input_ff memory is allocated first for post_activation and next post_activation is assigned the input, which is passed to function. But the allocated memory is not freed. Here the example from model.c:

void model_tensor_input_ff(model* m, int tensor_depth, int tensor_i, int tensor_j, float* input){
    if(m == NULL)
        return;
    int i,j,z,w,count,count2,z2,k1 = 0, k2 = 0, k3 = 0;
    /* Setting the input inside a convolutional structure*/
    cl* temp = (cl*)malloc(sizeof(cl));
    temp->post_activation = (float*)malloc(sizeof(float)*tensor_depth*tensor_i*tensor_j);
    temp->normalization_flag = NO_NORMALIZATION;
    temp->pooling_flag = NO_POOLING;
    temp->activation_flag = SIGMOID;
    temp->n_kernels = tensor_depth;
    temp->rows1 = tensor_i;
    temp->cols1 = tensor_j;
    temp->post_activation = input;
    temp->layer = -1;

Deleting the line:
temp->post_activation = (float*)malloc(sizeof(float)*tensor_depth*tensor_i*tensor_j);
solves the problem.

But maybe it causes some side effects?