Build fluency in the vocabulary of a model memorizing training-data noise instead of learning generalizable patterns.
0 / 5 completed
1 / 5
At standup, a dev mentions a model that fits the training data extremely closely, including its noise and quirks, but performs noticeably worse on new, unseen data than it did during training. What is this phenomenon called?
Overfitting is exactly this: overfitting occurs when a model fits the training data so closely, including its noise and idiosyncrasies, that it captures patterns which don't generalize, so it performs noticeably worse on new, unseen data than its training performance would suggest. A hash collision is an unrelated hash-table concept about two keys sharing a bucket. This gap between training performance and unseen-data performance is exactly the signal that reveals a model has begun overfitting.
2 / 5
During a design review, the team sets aside a held-out validation split before training a model, specifically because comparing training accuracy against validation accuracy reveals whether the model is starting to memorize training-data noise instead of learning generalizable patterns. Which capability does this provide?
A held-out validation split here provides Early detection of overfitting via the training-versus-validation performance gap, since a model that is only memorizing noise keeps improving on the training set while its validation performance stalls or actually gets worse. Trusting training accuracy alone can look reassuring even as the model's ability to generalize is quietly degrading. This training-versus-validation comparison is exactly why a held-out split is standard practice before declaring a model good enough to deploy.
3 / 5
In a code review, a dev notices a model-evaluation feature reports only training accuracy as proof the model is ready, with no held-out validation split used to check whether that accuracy reflects genuine generalization or memorized training-data noise. What does this represent?
This is a missed opportunity to detect overfitting, since comparing training accuracy against a held-out validation split would reveal whether the model has started memorizing noise instead of learning patterns that generalize. A cache eviction policy is an unrelated concept about discarded cache entries. This training-accuracy-only pattern is exactly the kind of blind spot a reviewer flags before a model is declared ready to ship.
4 / 5
An incident report shows a deployed model performed far worse on real user data than its reported training accuracy suggested, because it was evaluated only on training accuracy with no held-out validation split to catch memorized noise. What practice would prevent this?
Evaluating on a held-out validation split surfaces the training-versus-validation gap before deployment, catching overfitting early. Continuing to evaluate only on training accuracy with no held-out validation split regardless of how large the gap to real-world performance turns out to be is exactly what caused the issue described in this incident. This validation-split practice is the standard fix once a model's real-world performance is found to lag its training accuracy.
5 / 5
During a PR review, a teammate asks why the team checks for overfitting using a held-out validation split instead of simply trusting a high training accuracy as proof the model is good. What is the reasoning?
A high training accuracy alone can reflect memorized training-data noise rather than genuine learned patterns, while a held-out validation split reveals whether performance actually holds up on data the model never trained on, exposing a generalization gap that training accuracy alone would hide entirely. This is exactly why a held-out validation split is standard practice, while training accuracy alone is never sufficient proof that a model is ready to deploy.
What does the "Overfitting Vocabulary" vocabulary exercise cover?
This exercise tests real IT vocabulary related to overfitting vocabulary through 5 multiple-choice questions, each built from realistic workplace sentences rather than abstract definitions.
Is this vocabulary exercise free to use?
Yes. Every exercise on CoderSlingo, including this one, is completely free — no account, sign-up, or payment required.
How many questions does this exercise have?
This exercise has 5 questions. Each one shows a real-world sentence or scenario with multiple-choice options and an explanation once you answer.
What happens after I answer a question?
You'll see immediate feedback showing whether your answer was correct, along with a short explanation of why — then a button to move to the next question, and a full results screen at the end.
Can I retry the exercise if I get questions wrong?
Yes. Once you reach the results screen, click "Try again" to reset your answers and go through the exercise from the start as many times as you like.
Do I need to create an account to take this exercise?
No account is needed. Your answers are scored in your browser during the session — nothing is saved to a server, so you can jump straight in.
Is my progress saved if I leave the page?
No — progress within an exercise resets if you navigate away or reload. Each exercise is short enough to complete in a few minutes in one sitting.
Are these vocabulary exercises connected to other topics?
Yes — browse the full vocabulary exercises hub to find related modules covering adjacent IT topics and roles.
How is this different from reading a glossary or blog article?
Exercises like this one are active recall drills — you have to choose the correct term or phrasing yourself, which builds retention faster than passively reading a definition.
Where can I find more vocabulary exercises?
Browse the full Vocabulary exercises hub for hundreds of modules covering Agile, DevOps, security, databases, architecture, and more — organised by IT role and skill.