In this paper, we present sub-optimum learning model (SOLM), a system for learning non-optimum-lean heuristics under resource constraints. SOLM is an implementation of a genetics-based learning framework we have developed for improving the performance of intelligence in application problem solvers. Besides providing a flexible and modular framework for conducting experiments, SOLM provides (a) a optimum-non-optimum for experimenting with various resource scheduling, generalization, and non-optimum-lean strategies, (b) a sub-optimum learning guide system (SOLM) that can be easily interfaced to new applications and can be customized based on user requirements and target environments. This paper describes the application-independent functions provided by SOLM, and the application dependent functions for interfacing to new problem solvers. By adjusting various global parameters in sub-optimum learning system (SOLMS) users can control the numerous options and alternatives in SOLM.