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

# Logs

> Follow a run from another terminal. Default run detaches.

```bash theme={null}
compute logs <run_id>
compute logs <run_id> -f
```

`-f` follows until the run is terminal.

Without `--wait` on `compute run`, the CLI exits after create. Open another terminal and follow logs there.

```bash theme={null}
compute run train.py::main --gpu H100-SXM --yes
# note the run_id
compute logs <run_id> -f
```

`--wait` on `run` already streams status (and logs when present). Use `logs -f` when you detached.

`compute runs status <run_id>` is the state machine. `logs` is the stream.