You are evaluating two different language models, Model A and Model B, on a question-answering task. For a single training example, the correct answer span begins at token 15 and ends at token 18. The models produce the following probabilities for the correct start and end tokens:

• Model A: Probability of token 15 being the start = 0.8; Probability of token 18 being the end = 0.7.
• Model B: Probability of token 15 being the start = 0.5; Probability of token 18 being the end = 0.4.

Using the loss calculation for a single example, Loss = - (log p_start + log p_end), determine which model is performing better on this specific example. Justify your answer by calculating and comparing the loss for each model. (Use the natural logarithm, ln, for your calculations).