A long-standing limitation in developing functional soft materials with muscle-like properties is the challenge of achieving self-arranged structures. This study developed an anisotropic dual-network (DN) hydrogel via a training-induced strategy using zwitterionic-based composite materials. This bottom-up structural formation mechanism primarily relies on the asymmetric response of zwitterionic rigid and flexible chains to external mechanical stimuli, as well as the long-term memory effect of short chains in the rigid network from mechanical training. This simple mechanical training method significantly enhances the material's mechanical properties, including increasing the hydrogel's storage modulus by approximately threefold and making it resistant to cracks. These excellent mechanical properties make it highly promising for crack-resistant wearable sensors. Notably, leveraging its micron-scale directional cues, we also explored its potential as a substrate for directional cell growth.