The GitHub Actions job "Lint" on tvm.git/fix/onnx-integer-div-truncation has failed. Run started by GitHub user viiccwen (triggered by viiccwen).
Head commit for run: 966e8a2fea083516813a319590d91bc5f379f54e / viiccwen <[email protected]> [Fix][Relax][ONNX] Preserve integer Div truncation during import ONNX integer Div uses truncating division, rounding toward zero. The Relax ONNX frontend already special-cased integer Div to detect zero divisors, but its PrimExpr folding path could still use NumPy floating-point division and produce floating-point TIR values for integer shape computations. This can make shape-derived expressions such as Shape -> Gather -> Div -> Slice fail during import, because strided_slice expects integer PrimExpr bounds. Handle scalar integer Div inputs that contain a PrimExpr using TIR truncdiv, while keeping existing generic BinaryBase handling for tensor constants and runtime expressions. Add regression tests for integer constant Div semantics and a shape-derived PrimExpr Div used as a Slice bound. Signed-off-by: viiccwen <[email protected]> Report URL: https://github.com/apache/tvm/actions/runs/29073889445 With regards, GitHub Actions via GitBox --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
