[Relay/TOPI][ONNX/TFLite] Refactor MATRIX_SET_DIAG Operator for Relay… #9329
Conversation
…/TOPI to support ONNX Trilu operator
shengxinhu
requested review from
anijain2305,
areusch,
comaniac,
Huyuwei,
jcf94,
jroesch,
junrushao,
jwfromm,
kazum,
kevinthesun,
Laurawly,
MarisaKirisame,
masahi,
mbrookhart,
merrymercy,
siju-samuel,
slyubomirsky,
srkreddy1238,
tqchen,
vinx13,
wweic,
yzhliu,
zhiics and
ZihengJiang
as code owners
| // Determining which diagonal/sub-diagonal/super-diagonal it is | ||
| k = iter_vars[ndim] - iter_vars[ndim - 1]; | ||
| diagonal_indices.push_back(k2 - k); | ||
| diagonal_indices.push_back(k2(0) - k); |
There was a problem hiding this comment.
I believe (though please check) that you can just do k2() for 0-D tensor access.
There was a problem hiding this comment.
k2() will report error
python/tvm/relay/frontend/onnx.py
Outdated
| """Operator converter for Trilu""" | ||
|
|
||
| @classmethod | ||
| def _impl_v1(cls, inputs, attr, params): |
There was a problem hiding this comment.
this should be opset14 so _impl_v14
python/tvm/relay/frontend/onnx.py
Outdated
|
|
||
| @classmethod | ||
| def _impl_v1(cls, inputs, attr, params): | ||
| upper = attr.get("upper") |
There was a problem hiding this comment.
suggest attr.get("upper", 1) since this appears to be the default in onnx spec
|
onnx and onnxruntime version need to be updated to version 1.10.1 and 1.9.0 respectively for Trilu operator, |
|
Let me talk to some folks about doing this. |
|
Hmm this will eventually get merged, I might get to upgrading CI next week |
|
Working on upgrading onnx right now #9511 |
|
wait for #9882 |
|
The new image with updated ONNX should become available by this week. |
|
New CI image with updated onnx is (finally up)! Please jostle ci by sending an empty commit |
|
@shengxinhu do you still have interest in working to get this PR merged? If not then I can try to shepard this in new branch. |
Thanks, please shepard it. |
…/TOPI to support ONNX Trilu operator
This commit is based on PR apache#9329 proposed by @shengxinhu.
Refactor MATRIX_SET_DIAG operator in Relay/TOPI to support ONNX Trilu operator;
+ Fixed issues related to shape transformation of inputs in tflite and onnx frontend ops.
areusch
added
needs-triage
Refactor MATRIX_SET_DIAG operator in Relay/TOPI to support ONNX Trilu operator