Function load_model 

pub fn load_model<const IN: usize, const OUT: usize>(
    path: impl AsRef<Path>,
) -> Result<NeuralNetwork<IN, OUT>, ModelError>

Reads a model from path, checking that its shape matches IN and OUT.

Files written by every earlier version of this library load too. Every layer is checked against its neighbours, so a corrupted file is reported here rather than panicking later in NeuralNetwork::feed_forward, and the file must end where the model does.

Errors

Returns ModelError::Io if the file cannot be read, ModelError::UnrecognizedFormat if it is not a model, ModelError::UnsupportedVersion if a newer version of this library wrote it, ModelError::Malformed if it is truncated or corrupted, ModelError::DimensionMismatch if the stored network does not fit NeuralNetwork<IN, OUT>, and ModelError::EmptyNetwork or ModelError::InconsistentLayer if its layers do not form a network at all.

Example

let nn: NeuralNetwork<2, 1> = load_model("model.bin")?;