This paper investigates the training of quantum binary classifiers using quantum simulators with limited computational resources. It demonstrates that, although memory size remains the primary limitation for simulators, modern classical computing systems can effectively simulate quantum schemes, achieving classification accuracy comparable to classical models within fewer epochs in short time frames when constrained to a small number of qubits. Additionally, it is shown that training time scales exponentially with the number of qubits even when the number of weights does not change, although training time also scales linearly with the number of qubits.