tracel-ai/burn · error

bicubic interpolation backward is not supported for ndarray

Error message

bicubic interpolation backward is not supported for ndarray backend

What it means

interpolate_backward in the ndarray backend has no backward implementation for InterpolateMode::Bicubic and panics. Bicubic resizing gradients are simply not implemented for this CPU backend. Only Nearest mode backward is supported.

Source

Thrown at crates/burn-ndarray/src/ops/module.rs:359

    fn interpolate_backward(
        x: FloatTensor<Self>,
        grad: FloatTensor<Self>,
        output_size: [usize; 2],
        options: InterpolateOptions,
    ) -> FloatTensor<Self> {
        match options.mode {
            InterpolateMode::Nearest => module_op!(inp(x, grad), opt(), E, |x, grad| {
                nearest_interpolate_backward::<E>(x, grad, output_size).into()
            }),
            InterpolateMode::NearestExact => {
                panic!("nearest exact interpolation backward is not supported for ndarray backend")
            }
            InterpolateMode::Bilinear => {
                panic!("bilinear interpolation backward is not supported for ndarray backend")
            }
            InterpolateMode::Bicubic => {
                panic!("bicubic interpolation backward is not supported for ndarray backend")
            }
            InterpolateMode::Lanczos3 => {
                panic!("lanczos3 interpolation backward is not supported for ndarray backend")
            }
        }
    }

    fn conv3d(
        x: FloatTensor<Self>,
        weight: FloatTensor<Self>,
        bias: Option<FloatTensor<Self>>,
        options: ConvOptions<3>,
    ) -> FloatTensor<Self> {
        module_op!(inp(x, weight), opt(bias), E, |x, weight, bias| conv3d::<E>(
            x, weight, bias, options
        )
        .into())
    }

View on GitHub (pinned to d16f7ba2ed)

Solutions

1. Use InterpolateMode::Nearest, which has a backward pass on ndarray
2. Switch backends (burn-cube/cubecl or burn-torch) which support bicubic backward
3. Keep bicubic only in inference paths outside the autodiff graph

if mode == InterpolateMode::Bicubic && backend_is_ndarray() { /* reconfigure */ }

fn supports_backward(mode: &InterpolateMode) -> bool {
    matches!(mode, InterpolateMode::Nearest)
}

When it happens

Trigger: Backpropagating through an interpolate node with InterpolateMode::Bicubic while using burn-ndarray as the backend.

Common situations: Training image super-resolution or generative models that use bicubic upsampling layers on the ndarray CPU backend.

Related errors

• bilinear interpolation backward is not supported for ndarray
• lanczos3 interpolation backward is not supported for ndarray
• nearest exact interpolation backward is not supported for nd
• Can't differentiate avg pool 2d backward.
• Can't differentiate max pool2d with indices backward.

AI-assisted analysis of tracel-ai/burn@d16f7ba2ed (2026-09-05). Data as JSON: /api/errors/93aa682cedae81d8. Report an issue: GitHub.