Temperature Decay Scheduler
Backpropagation, Training & Optimization DS practice problem on Onlearn.
Difficulty: medium.
Topics: Understanding Temperature Decay in Simulated Annealing and Softmax Scaling, Exponential Cooling Schedule, Temperature Floor Constraints, Geometric Decay Sequences, State Space Exploration, Boltzmann Distribution Control, Deep Learning Optimization, Probabilistic Modeling, Hyperparameter Tuning, Numerical Analysis, Stochastic Processes, Simulated Annealing, Softmax Temperature Scaling, Learning Rate Scheduling, Exploration vs Exploitation, Convergence Control.
Implement a 'TemperatureScheduler' class that manages the cooling schedule for a training process. The class should be initialized with an initial temperature, a decay rate, and a minimum temperature threshold. Implement a 'get temperature(step)' method that returns the current temperature based on an exponential decay formula: T(t) = max(T min, T 0 (decay rate ^ step)).