onnxruntime/onnxruntime
Tianlei Wu fb11c67368
Fix SkipLayerNorm for 2D input (#17014)
Fix an obvious bug:
(1) In packing mode, the input for SLN has two dimensions (introduced by
#15283): [token_count, hidden_size]. Current code of `element_count =
input_dims[0] * sequence_length * hidden_size` will use element_size =
token_count * hidden_size * hidden_size, and causes invalid memory write
in cuda kernel and ORT crash

and two minor issues:
(2) potential integer overflow in `static_cast<int>(element_count)`
(3) some dead code after `return LaunchSkipLayerNormKernel` that will
never have chance to run.
2023-08-08 14:04:03 -07:00
..
contrib_ops Fix SkipLayerNorm for 2D input (#17014) 2023-08-08 14:04:03 -07:00
core Add PerThreadContext for TRT EP (#16599) 2023-08-08 13:02:34 -07:00
python Updating QDQ to support Float8E4M3FN (#16550) 2023-08-08 12:18:48 +02:00
test Add PerThreadContext for TRT EP (#16599) 2023-08-08 13:02:34 -07:00
tool/etw
wasm [js/web] enable ONNX Runtime Web error messages in JS (#16335) 2023-06-15 09:45:41 -07:00
__init__.py ExecutionProvider API refactor - move allocator from EP level to SessionState level and indexed by OrtDevice (#15833) 2023-06-19 17:44:45 -07:00
ReformatSource.ps1
ReformatSourcePython.bat
VSCodeCoverage.runsettings