> For the complete documentation index, see [llms.txt](https://www.mlcompendium.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://www.mlcompendium.com/evals/intro.md). # Introduction A model is not a result until it has a score, a split, and a baseline. After this chapter the reader can choose a metric, a validation scheme, and a test that would catch a broken dataset, and calibration, explanation, and fairness wait in Responsible AI. The score comes first. [Evaluation Metrics](/evals/evaluation-metrics.md) collects the supervised and unsupervised measures, from accuracy and precision/recall through ROC, F1, and perplexity, with a reality check on the metric-learning literature. A score is only as honest as the data under it, so [Datasets Reliability & Correctness](/evals/validation.md) is about why we shouldn't trust models: dataset ablation, shortcut cues, and behavioral testing that ask whether a dataset actually requires the reasoning you think it does. Once the score and the data can be trusted, the model has to keep earning them. [Training Strategies](/evals/training-strategies.md) is the framework for a continuous training strategy, with periodic and performance-based retraining driven by data changes, window size, and what to retrain. [Hyper Parameter Optimization](/evals/hyper-parameter-optimization.md) is searching hyperparameters with Hyperopt, HyperparameterHunter, and a comparison of optimizers, without losing track of experiments. The last two pages turn evaluation into something a pipeline runs and a field compares. [Data & Model Tests](/evals/data-and-model-tests.md) validates data in ML pipelines and unit tests models, from Great Expectations-style data checks to pytest and mocks. [Benchmarking](/evals/benchmarking.md) collects benchmarks across datasets, algorithms, hardware, cloud providers, and platforms, including NLP and multi-task learning leaderboards and notes on scaling networks. All of it rests on the split being real. [Data Scientists — Stop Using A Random Seed](https://cohenori.medium.com/data-scientists-stop-using-a-random-seed-fd8f2b83ed50) (April 2022) is why a score needs a real split. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://www.mlcompendium.com/evals/intro.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.