tracel-ai/burn · error

Failed to load record

Error message

Failed to load record

What it means

load_record is the infallible wrapper around try_load_record; it panics with 'Failed to load record' when the record does not match the module's structure (missing/extra fields, shape or dtype mismatches). The library panics because record validation failures are considered programming errors rather than runtime conditions.

Source

Thrown at crates/burn-core/src/module/base.rs:482

        self,
        record: crate::store::ModuleRecord,
    ) -> Result<Self, crate::store::RecordError>
    where
        Self: Sized,
    {
        record.apply(self)
    }

    /// Apply a [`ModuleRecord`](crate::store::ModuleRecord) to this module, consuming and returning
    /// it.
    ///
    /// Panics if validation fails; use [`try_load_record`](Module::try_load_record) for the
    /// fallible variant.
    fn load_record(self, record: crate::store::ModuleRecord) -> Self
    where
        Self: Sized,
    {
        self.try_load_record(record).expect("Failed to load record")
    }

    /// Save this module's parameters to a burnpack file on disk.
    ///
    /// Convenience for [`into_record`](Module::into_record) followed by
    /// [`ModuleRecord::save`](crate::store::ModuleRecord::save). For non-default load behavior
    /// (dtype policy, partial loading, validation), go through the record directly.
    #[cfg(feature = "std")]
    fn save_file<P: AsRef<std::path::Path>>(self, path: P) -> Result<(), crate::store::RecordError>
    where
        Self: Sized,
    {
        self.into_record().save(path)
    }

    /// Load this module's parameters from a burnpack file on disk, returning the loaded module.
    ///
    /// Uses the default load behavior. Panics on I/O or validation errors; use

View on GitHub (pinned to d16f7ba2ed)

Solutions

1. Switch to try_load_record and handle the Result to inspect the exact validation error.
2. Regenerate the record/checkpoint from the current module definition.
3. Ensure the module architecture (shapes, dtypes, names) matches the one that saved the record.

Example fix

// before
let model = Model::load_record(record);
// after
let model = Model::try_load_record(record)
    .expect("checkpoint mismatch: regenerate the record for the current model");

// Verify record compatibility before loading (fallible variant)
let model = match Model::try_load_record(record) {
    Ok(m) => m,
    Err(e) => panic!("record incompatible with model: {e}"),
};

let model = Model::try_load_record(record)
    .unwrap_or_else(|e| panic!("Failed to load record: {e}"));

When it happens

Trigger: Calling module.load_record(record) where the record was saved from a module whose parameter shapes, dtypes, or field names differ from the target module.

Common situations: Loading checkpoints saved from an older model version after changing layer sizes; loading a checkpoint into a differently-configured model; record serialization format changes between burn versions.

Related errors

• Should match at least one parameter group.
• Optimizer record tensors should carry a parameter id.
• burn-store: a panic escaped Module::map during ModuleSnapsho
• Failed to load module from file
• Can delete model checkpoint.

AI-assisted analysis of tracel-ai/burn@d16f7ba2ed (2026-09-05). Data as JSON: /api/errors/0721af9ce0389165. Report an issue: GitHub.