tracel-ai/burn · error

int_scatter with {other:?} update is not implemented

Error message

int_scatter with {other:?} update is not implemented

What it means

int_scatter in burn-tch handles only IndexingUpdateOp::Add and IndexingUpdateOp::Mul update operations via TchOps::scatter/scatter_mul. Any other IndexingUpdateOp variant passed in falls through to unimplemented!() and panics, naming the unsupported op.

Source

Thrown at crates/burn-tch/src/ops/int_tensor.rs:303

    fn int_scatter(
        dim: usize,
        tensor: TchTensor,
        indices: TchTensor,
        value: TchTensor,
        update: burn_backend::tensor::IndexingUpdateOp,
    ) -> TchTensor {
        match update {
            burn_backend::tensor::IndexingUpdateOp::Assign => {
                TchOps::scatter_assign(dim, tensor, indices, value)
            }
            burn_backend::tensor::IndexingUpdateOp::Add => {
                TchOps::scatter(dim, tensor, indices, value)
            }
            burn_backend::tensor::IndexingUpdateOp::Mul => {
                TchOps::scatter_mul(dim, tensor, indices, value)
            }
            other => unimplemented!("int_scatter with {other:?} update is not implemented"),
        }
    }

    fn int_scatter_nd(
        data: TchTensor,
        indices: TchTensor,
        values: TchTensor,
        reduction: burn_backend::tensor::IndexingUpdateOp,
    ) -> TchTensor {
        TchOps::scatter_nd(data, indices, values, reduction)
    }

    fn int_gather_nd(data: TchTensor, indices: TchTensor) -> TchTensor {
        TchOps::gather_nd(data, indices)
    }

    fn int_select(tensor: TchTensor, dim: usize, indices: TchTensor) -> TchTensor {
        TchOps::index_select_dim(tensor, dim, indices)

View on GitHub (pinned to d16f7ba2ed)

Solutions

1. Use the supported Add or Mul update ops for scatter on LibTorch
2. Emulate the desired update manually: mask_where / select_assign with arithmetic built from supported ops
3. Implement the missing variant with existing torch scatter primitives (e.g. scatter_add on negated values for Sub)
4. Upgrade burn to a version that maps the op you need

Example fix

// before
tensor.scatter(dim, indices, value, IndexingUpdateOp::Sub);

// after
tensor.scatter(dim, indices, -value, IndexingUpdateOp::Add); // emulate Sub via Add of negation

match op {
    IndexingUpdateOp::Add | IndexingUpdateOp::Mul => { /* safe */ }
    other => panic!("unsupported op {:?}, rewrite as Add/Mul", other),
}

When it happens

Trigger: Calling scatter on an integer tensor with a LibTorch backend using an update op other than Add or Mul (e.g. a Sub/Max/Min variant of IndexingUpdateOp).

Common situations: Using higher-level indexing/assign APIs where the update op is configurable; code ported from a backend that supports more scatter update ops; Burn upgrades introducing new IndexingUpdateOp variants not yet wired into tch.

Related errors

• Unsupported dtype for `bool_from_data`
• Unsupported dtype for `int_from_data`: {:?}
• int_select_assign with {other:?} update is not implemented
• Operation not marked for autodiff.
• Cannot move a tensor from a capture device

AI-assisted analysis of tracel-ai/burn@d16f7ba2ed (2026-09-05). Data as JSON: /api/errors/7f4fe28de4f9990b. Report an issue: GitHub.