Hi, I have a DLTensor which malloc like this:
int dtype_code = kDLFloat;
int dtype_bits_16 = 16;
int dtype_lanes = 1;
int device_type = kDLCPU;
int device_id = 0;
int64_t shape_coff[2] = {batch, 4};
TVMArrayAlloc(shape_coff, 2, dtype_code, dtype_bits_16, dtype_lanes,
device_type, device_id, &coeff);
How should I write data into this memory field? My data is calculate from float, if using float assign values, the values readout is wrong.
I am confused on how to write data to a fp16 Tensor? Does anybody knows how?