> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trymonte.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> Ways to score a model

An eval scores a model on a fixed set of tasks with a grader. In Monte, an eval is an [eval environment](https://docs.trymonte.ai/environments/overview.md): task rows and a grader, added with `monte env add --eval`. The team's own tasks and a public benchmark's tasks become an eval the same way.

## Ways to Evaluate

| Method | Scores | Recorded | Page |
| - | - | - | - |
| Eval environment in an experiment | A base model or a checkpoint, on pinned tasks | Yes, as a change from the baseline | [Scoring Models](https://docs.trymonte.ai/evaluation/scoring-models.md) |
| Regrade | The saved replies of an eval run, with another grader | Yes, in another experiment | [Regrade a Run](https://docs.trymonte.ai/evaluation/scoring-models.md#regrade-a-run) |
| Eval harness | A deployment, through its endpoint | No | [Use an Eval Harness](https://docs.trymonte.ai/serving/endpoints.md#use-an-eval-harness) |

Only the first two produce numbers that Monte compares. Use them to choose a checkpoint.

## What an Eval Run Records

* **A score**: the mean task score, with a 95% interval
* **The change** from the baseline: the paired mean difference of task scores, and the tasks fixed and broken at the eval's pass threshold
* **A [trace](https://docs.trymonte.ai/observability/traces.md)** for each task: the prompt, the reply, the score, and the expected answer
* **The settings** and the exact environment version, by fingerprint

## Evaluate a Checkpoint

1. [Create an experiment](https://docs.trymonte.ai/evaluation/experiments.md#create-an-experiment) with held-out and development evals.
2. [Take the baselines](https://docs.trymonte.ai/evaluation/experiments.md#take-the-baselines) on the base model.
3. Train, then [score each checkpoint](https://docs.trymonte.ai/evaluation/scoring-models.md#score-a-checkpoint) on the development evals.
4. [Choose a checkpoint](https://docs.trymonte.ai/evaluation/scoring-models.md#choose-a-checkpoint) and score it once on the held-out eval.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.