Negative Log-Likelihood
A fundamental loss measure in language model pretraining and evaluation, equal to the negative logarithm of the model's predicted probability for the correct answer. A lower NLL indicates more confident and accurate model predictions.
Timeline (last 90 days)
The coefficient of variation (standard deviation divided by mean) is a dimensionless normalized measure that eliminates the influence of differences in the absolute NLL levels between items.
Under the assumption of a Gaussian output distribution, minimizing negative log-likelihood (NLL) and minimizing mean squared error (MSE) are mathematically completely equivalent.
All Facts (2)
The coefficient of variation (standard deviation divided by mean) is a dimensionless normalized measure that eliminates the influence of differences in the absolute NLL levels between items.
50%UnverifiedUnder the assumption of a Gaussian output distribution, minimizing negative log-likelihood (NLL) and minimizing mean squared error (MSE) are mathematically completely equivalent.
50%