Possible typo in docs\OpperatorKernels.md

[georgen117]

I was looking at the recent work to to add LongformerAttention to the code and found Longformer

https://github.com/microsoft/onnxruntime/blob/master/docs/OperatorKernels.md

Op Name	Parameters	OpSet Version	Types Supported
LongformerAttention	in input:T in weight:T in bias:T in mask:T in global_weight:T in global_bias:T in global:G out output:T	1+	T = tensor(float), tensor(float16)

The global:G seems to be a typo should the G be a T?

Since the top of the document indicates the file is automatically generated I was wondering if the G is indeed a typo does it exist some place else?

[ytaous]

doesn't look like a typo, it's just type alias, see:
https://github.com/microsoft/onnxruntime/blob/master/docs/ContribOperators.md#com.microsoft.LongformerAttention

[georgen117]

doesn't look like a typo, it's just type alias, see: https://github.com/microsoft/onnxruntime/blob/master/docs/ContribOperators.md#com.microsoft.LongformerAttention

Agreed: G is its own type.

Type Constraints
T : tensor(float), tensor(float16)
Constrain input and output types to float tensors.
G : tensor(int32)
Constrain to integer types

I think where I was confused was that 'G' was not in the Types supported column. It should probably be updated. So I guess there is still a typo of sorts. It is lacking one of the "Type Supported" values