# Reading a failed generation (/docs/troubleshooting/failed-generation)





When a generation fails, SimpliGen shows the reason on the job in the [Queue](/docs/generating/queue-gallery). Start there.

## First steps [#first-steps]

1. Click the **Queue*&#x2A; button on the right rail to open the Queue drawer, then find the failed job under &#x2A;*"Needs attention"**.
2. Read the error message shown on the card. It usually points at the cause.
3. Try the matching fix below.

<img alt="A failed generation card in the Queue" src="__img0" />

## Common causes and fixes [#common-causes-and-fixes]

**Out of memory, or stuck near the end (local video).**
The GPU ran out of memory, or the final step fell back to the CPU. Use a lower resolution or shorter clip, or generate in the cloud. See [Out of memory or stuck generations](/docs/troubleshooting/out-of-memory).

**A model or dependency is missing or broken (local).**
Click the **Local Engine*&#x2A; item on the rail's bottom system strip, choose &#x2A;*"Manage engine"*&#x2A;, and use &#x2A;*"Reinstall Dependencies"** on the affected pack, or reinstall the preset's models. See [Engine install problems](/docs/troubleshooting/engine-install).

**Network or cloud error.**
Check your connection and try again. If your cloud session dropped, reconnect. See [Managing your cloud connection](/docs/cloud-credits/managing-connection).

**Invalid input.**
For example a value the model does not accept. Reset the setting in question to its default and regenerate.

## Cloud failures are refunded [#cloud-failures-are-refunded]

If a cloud generation fails or produces nothing, the credits are refunded automatically. If you were charged for a result you never received, contact support on Discord.

## Still stuck [#still-stuck]

If the error is unclear or keeps happening, export your logs from **Settings** and share them with support on Discord. The logs contain the detail needed to diagnose it.
