LESSON 2.2 · Derivation · 90 min
Negative log-likelihood: the model scorecard
Negative log-likelihood: the model scorecard. Learn likelihood, logarithms, NLL, and cross-entropy and complete: Calculate mean NLL from a probability table. Part of the “Language
Learning objectives
- Explain what problem “Negative log-likelihood: the model scorecard” solves without hiding behind terminology.
- Trace the variables and causal links across likelihood, logarithms, NLL.
- Complete “Calculate mean NLL from a probability table” and judge the result with evidence rather than intuition.
Core concepts
likelihood
likelihood is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment before moving to an optimized implementation.
logarithms
logarithms is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.
NLL
NLL is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.
and cross-entropy
and cross-entropy is part of the lesson’s causal model. State its inputs, outputs, invariants, and failure mode; then verify it with a hand-check or a minimal experiment.
Build and verify
Calculate mean NLL from a probability table
- Predict: write the expected output, trend, or failure before running code.
- Build: implement only the minimum components needed to answer the question.
- Verify: compare with a baseline or trusted implementation; save seeds, parameters, and raw outputs.
- Transfer: change one shape, dataset, scale, or workload condition and explain whether the conclusion still holds.