1 Overview | 1.1 Why use the Maximum Entropy (MaxEnt) model? | 1.2 OT: A quick recap | 1.3 The MaxEnt model defined | 1.4 Calculating MaxEnt probabilities: Step-by-step | 1.4.1 Number of violations, $f_k(SR_i, UR_i)$ | 1.4.2 Penalty | 1.4.3 MaxEnt score | 1.4.4 Normalization, $Z$ | 1.4.5 Predicted probability | 1.5 Characteristics of the MaxEnt model | 2 Fitting the model (basic): optimize_weights() | 2.1 Preparing the data file | 2.2 Finding weights to fit the model: No bias | 2.2.1 Retrieving specific model attributes | 2.3 Fitting the model to data: Explained | 2.3.1 Defining log likelihood | 2.3.2 What guides the weight changes?: Maximizing log likelihood | 3 Fitting the model with priors: optimize_weights() | 3.1 Unpacking the math behind the bias term | 3.2 Using the prior: Which argument to use? | 3.3 Overfitting & regularization | 3.3.1 Ceiling log likelihood | 3.3.2 Regularization as a strategy to combat overfitting | 3.4 Incorporating substantive effects | 3.4.1 The phenomenon: Productivity of saltation (White 2017) | 3.4.2 Modeling the extension of p→v (trained) to b→v (untrained) | 3.4.3 Modeling the extension of b→v (trained) to p→v (untrained) | 3.4.4 Conclusion | 4 Making predictions: predict_probabilities() | 4.1 Get predictions with trained weights | 4.2 Get predictions with User-chosen weights | 4.3 Save predictions to file | 5 The full MaxEnt model: With Temperature | 5.1 The full MaxEnt model defined | 5.2 The effect of temperature on probability | 5.3 Using temperature: Experimental data from "2 alternative forced choice" tasks | 5.3.1 Hungarian backness harmony | 5.3.2 Challenges with assessing under-/over-learning | 5.3.3 A rigorous approach to assessing under-/over-learning | 5.3.4 Generating depolarized predictions for a 2AFC task | 5.3.5 Monte-carlo simulations based on the predicted probabilities above | 5.3.6 Are unnatural constraints under-learned? | 6 Comparing models: compare_models() | 6.1 Likelihood ratio test | 6.1.1 Likelihood ratio test: Background | 6.1.2 Likelihood ratio test: Explained | 6.1.3 Likelihood ratio won't work for non-nested models | 6.2 Akaike Information Criterion (AIC) | 6.2.1 AIC: Background | 6.2.2 Compare models using the AIC method | 6.2.3 AIC weight (aic.wt) explained | 6.2.3.1 AIC score distance, $\delta$ (aic.delta) | 6.2.3.2 Raw AIC weight, $r$ | 6.2.3.3 Actual AIC weight: aic.wt | 6.3 Bayesian Information Criterion (BIC) | 6.3.1 BIC: Background | 6.3.2 Compare models using the BIC method | 6.4 Akaike Information Criterion corrected for small sample size (AIC-C) | 6.4.1 AIC-C: Background | 6.4.2 Compare models using the AIC-C method | 6.4.3 A potential pitfall