tracel-ai/burn · error

Can delete policy checkpoint.

Error message

Can delete policy checkpoint.

What it means

Guard in the RL checkpointer's `checkpoint`: deleting an old policy checkpoint for the epoch selected by the checkpointing strategy failed (I/O or missing file), so the loop panics rather than continue with possibly unbounded checkpoint accumulation.

Source

Thrown at crates/burn-train/src/learner/rl/checkpointer.rs:39

    RLPolicyRecord<RLC>: Checkpoint,
    RLAgentRecord<RLC>: Checkpoint,
{
    /// Create checkpoint for the training process.
    pub fn checkpoint(
        &mut self,
        policy: &RLC::PolicyState,
        learning_agent: &RLC::LearningAgent,
        epoch: usize,
        store: &EventStoreClient,
    ) {
        let actions = self.strategy.checkpointing(epoch, store);

        for action in actions {
            match action {
                CheckpointingAction::Delete(epoch) => {
                    self.policy
                        .delete(epoch)
                        .expect("Can delete policy checkpoint.");
                    self.learning_agent
                        .delete(epoch)
                        .expect("Can delete learning agent checkpoint.")
                }
                CheckpointingAction::Save => {
                    self.policy
                        .save(epoch, policy.clone().into_record())
                        .expect("Can save policy checkpoint.");
                    self.learning_agent
                        .save(epoch, learning_agent.record())
                        .expect("Can save learning agent checkpoint.");
                }
            }
        }
    }

    /// Load a training checkpoint.
    pub fn load_checkpoint(

View on GitHub (pinned to d16f7ba2ed)

Solutions

1. Check permissions and disk state of the checkpoint directory
2. Inspect the underlying error from the record deletion
3. Handle stale/missing epochs gracefully if concurrent runs share the directory

When it happens

Trigger: Thrown at crates/burn-train/src/learner/rl/checkpointer.rs:39 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/59b300cbd5c080dc. Report an issue: GitHub.