tracel-ai/burn · error

A minimum of one device.

Error message

A minimum of one device.

What it means

Configuration guard in the multi-device training strategy's `run_optim_main`: the devices list used for distributed training is empty, so no work can be distributed and execution panics; at least one device (the main device) is required.

Source

Thrown at crates/burn-train/src/learner/supervised/strategies/multi/epoch.rs:89

            "Executing training step for epoch {} on devices {:?}",
            epoch,
            devices
        );

        let mut iterators = self
            .dataloaders
            .iter()
            .map(|d| d.iter())
            .collect::<Vec<_>>();
        let mut iteration = 0;
        let mut accumulator = GradientsAccumulator::new();
        let mut accumulation_current = 0;

        let accumulation = self.grad_accumulation.unwrap_or(1);
        let step = MultiDevicesTrainStep::<M>::new(&devices);

        // The main device is always the first in the list.
        let device_main = devices.first().expect("A minimum of one device.").clone();

        loop {
            let (items, progress) = match step.step(iterators.as_mut_slice(), &learner.model()) {
                Ok(result) => result,
                Err(err) => {
                    interrupter.stop(Some(&format!("dataset error during training step: {err}")));
                    break;
                }
            };
            if items.is_empty() {
                break;
            }

            learner.lr_step();

            let mut progress_items = Vec::with_capacity(items.len());
            for item in items.into_iter() {
                let grads = item.output.grads.to_device(&device_main, &learner.model());

View on GitHub (pinned to d16f7ba2ed)

Solutions

1. Pass at least one device to the multi-device learner configuration
2. Verify device enumeration returned a non-empty list
3. Fall back to the single-device strategy if only CPU/one device is available

When it happens

Trigger: Thrown at crates/burn-train/src/learner/supervised/strategies/multi/epoch.rs:89 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of tracel-ai/burn@d16f7ba2ed (2026-09-05). Data as JSON: /api/errors/c4a162c823acc914. Report an issue: GitHub.