Abstract: Systems and methods quantize an application having a trained Deep Neural Network (DNN) for deployment on target hardware. The application may be instrumented to observe data values generated during execution of the application. Statistics may be generated for the observed data values and presented in a visualization tool. The application may be quantized through a rules based approach. The quantization may be based on the statistics and on constraints imposed by resources available at the target hardware. The systems and methods may present the proposed data types resulting from the quantization and may create a quantized version of the application incorporating the proposed data types. The systems and methods may generate performance data to validate the quantized version of the application. Changes to the rules may be made and the quantization process repeated if the performance is not satisfactory.

Abstract: A device may obtain program code, determine variables based on the program code, execute or analyze the program code, and determine values of the variables based on the executing or analyzing. The device may determine a range for each variable based on positions of at least one significant digit for the values. The range may be determined based on a count of a quantity of occurrences of the at least one significant digit for the values. The device may determine relationships between the variables based on the executing or analyzing, and may determine one or more effects on the range for the variables based on the relationships between the variables. The device may generate a visualization that presents the range for the variables, the relationships between the variables, and the one or more effects on the range for the variables, and may provide the visualization for display.