> ## 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.

# Scoring Models

> Score checkpoints and read the change

`monte eval` scores one model on one eval of an [experiment](https://docs.trymonte.ai/evaluation/experiments.md) and records an eval run.

## Score a Checkpoint

```bash theme={null}
monte eval support-triage support-escalation-dev --checkpoint train-07/200 --remote
```

The second argument names the eval. It can be left out only when the experiment has one development eval.

| Target | Flag |
| - | - |
| A base model | `--model Qwen/Qwen3-8B` |
| One checkpoint | `--checkpoint train-07/200` |
| The checkpoint at one step | `--step 150` |
| The newest checkpoint | No flag |

`--remote` runs the eval on a GPU instance. `--smoke` scores 3 tasks to test the setup and never counts as a result. [Eval Commands](https://docs.trymonte.ai/commands/evaluations.md#monte-eval) lists every flag.

## Follow, Inspect, or Stop a Run

```bash theme={null}
monte run logs support-triage/eval-03 -f
monte run list support-triage --kind eval
monte run show support-triage/eval-03
monte run stop support-triage/eval-03
```

`monte run list` prints each run with its alias, such as `eval@200`, and its score. `monte run show` prints the target, the settings, and the score of one run. `monte trace list` prints the per-task results. [Observability Commands](https://docs.trymonte.ai/commands/observability.md) lists every flag.

## Eval Settings

Each eval keeps the settings pinned when the experiment was created. An eval environment can declare its own values. Others take the default.

| Setting | Default |
| - | - |
| `temperature` | `0.0` (greedy) |
| `top_p` | `1.0` |
| `num_repeats` | `1` |
| `max_output_tokens` | The server default |
| `pass_threshold` | `1.0` |

No flag changes them. Other settings need a new experiment.

## Read the Result

```bash theme={null}
monte experiment show support-triage
```

```text theme={null}
  support-escalation-dev: 0.50 -> 0.60 (delta +0.10 ± 0.06 paired, 30 fixed / 10 broken, eval@200)
```

Each line shows the baseline, the newest score, and the change. A score is the mean task score. The change is the paired mean difference of task scores, with its 95% interval. **Fixed** tasks were below the eval's pass threshold at the baseline and reach it now. **Broken** tasks reached it and fall below it now. A change is clear when it stays above zero after the interval is taken off. The Console shows each run under **Evaluations**, with a [trace](https://docs.trymonte.ai/observability/traces.md) for each task.

> [accordion] How the change is computed
>
> Both runs score the same `n` tasks, so Monte compares them task by task. `d` is the list of per-task differences: the newest score minus the baseline score.
>
> ```text theme={null}
> change   = mean(d)
> interval = 1.96 * sd(d) / sqrt(n)
> ```
>
> The interval is 95%. Monte compares only runs on the same eval, in the same experiment, from the same base model.

## Regrade a Run

`monte grade` scores the replies of an eval run again with the grader of another experiment, without serving the model:

```bash theme={null}
monte grade support-triage/eval-03 --into support-triage-v2
```

[Fix a Grader and Regrade](https://docs.trymonte.ai/environments/graders.md#fix-a-grader-and-regrade) shows when to use it.

## Choose a Checkpoint

The best development change over many checkpoints overstates the gain, so the held-out score is the result to report.

1. Choose the checkpoint with the best development change.

2. Score the held-out eval on it:

   ```bash theme={null}
   monte eval support-triage support-escalation --checkpoint train-07/200
   ```

3. [Deploy](https://docs.trymonte.ai/serving/deploy.md) it:

   ```bash theme={null}
   monte deployment create support-triage/train-07/200 --name support-triage
   ```


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