Quantum Adaptive Self-Attention for Quantum Transformer Models