ZZY 0728 Helper method convert output to floatbuffer

Helper method to convert Output tensors to float and write them out to files.

Helper method to convert Output tensors to float and write them out to files.

Convert data to float or de-quantization. This is used when user requests for float output and the model produces non-float output.

Convert data to float or de-quantization. This is used when user requests for float output and the model produces non-float output.

Some existing graphs seem to be sensitive to modifications of int32_to_float So this is added in order to work around that

Some existing graphs seem to be sensitive to modifications of int32_to_float So this is added in order to work around that

Helper method to copy a float buffer, quantize it, and copy it to a tensor (Qnn_Tensor_t) buffer.

Helper method to copy a float buffer, quantize it, and copy it to a tensor (Qnn_Tensor_t) buffer.

--------------------------------------------------------------------------- pal::FileOp::copyOverFile ------------------------------------------------

--------------------------------------------------------------------------- pal::FileOp::copyOverFile ------------------------------------------------

copy from another hashmap. this is used in copy-ctor; and in operator = after 'clear'. We assume the current map is empty (no deleted items) but the t

copy from another hashmap. this is used in copy-ctor; and in operator = after 'clear'. We assume the current map is empty (no deleted items) but the t

@brief process invoke the make_op_custom function */

@brief process invoke the add_package_opt function */

@brief process invoke the make_op_custom function */

@brief process invoke the add_package_opt function */

------------------------------------------------------------------------------ pal::Directory::create ------------------------------------------------

The amount of space left in the current chunk, approximately. DO NOT CALL unless chunk_count() > 0

The amount of space left in the current chunk, approximately. DO NOT CALL unless chunk_count() > 0

for direct access to bulk_data, in contiguous tensors only data_ptr() can be assigned to.

Some methods return the same thing as in LayoutTensor, but with the type being element_type instead of storage_type.

for direct access to bulk_data, in contiguous tensors only data_ptr() can be assigned to.

Some methods return the same thing as in LayoutTensor, but with the type being element_type instead of storage_type.

--------------------------------------------------------------------------- pal::FileOp::deleteFile --------------------------------------------------