Online beat tracking (OBT) has always been a challenging task. Dueto the inaccessibility of future data and the need to make inferencein real-time. We propose Don’t Look back! (DLB), a novel approachoptimized for efficiency when performing OBT. DLB feeds theactivations of a unidirectional RNN into an enhanced Monte-Carlolocalization model to infer beat positions. Most preexisting OBTmethods either apply some offline approaches to a moving windowcontaining past data to make predictions about future beat positionsor must be primed with past data at startup to initialize. Meanwhile,our proposed method only uses activation of the current time frameto infer beat positions. As such, without waiting at the beginning toreceive a chunk, it provides an immediate beat tracking response,which is critical for many OBT applications. DLB significantlyimproves beat tracking accuracy over state-of-the-art OBT methods,yielding a similar performance to offline methods.