> For the complete documentation index, see [llms.txt](https://ai-ops-inc.gitbook.io/ai-ops-tank-sim-users-guide/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://ai-ops-inc.gitbook.io/ai-ops-tank-sim-users-guide/step-4.md).

# Step 4

Validating the Trained Agent Control Capability

## Overview

In Step 4, the reinforcement learning model has been produced, and the validation of the model will now be completed.  The agent validation will use both the Digital Twin built during [Step 2](/ai-ops-tank-sim-users-guide/step-2.md), and the trained agent model from [Step 3](/ai-ops-tank-sim-users-guide/step-3.md) to bench test the agent.  This step will produce a graph that will show what the actual system did during this time frame and how the agent would have preformed and the system would have responded to the agent.  From one graph, you can see what is, and what it can be.

<figure><img src="https://232492887-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F4LDEx85mFpnzJbFgrBc4%2Fuploads%2F7yUDVkRLN7yiqSjyHNoO%2FDQN.JPG?alt=media&amp;token=35d87350-653c-49ef-aab3-9dac44ed3312" alt=""><figcaption><p>Example DQN Validation where the Agent control is more stable and responsive</p></figcaption></figure>