๐โจ๐ก,๐กโฒโฉ is then used when computing the attention ๐โจ๐ก,๐กโฒโฉ that ๐ฆโจ๐กโฉ should pay to ๐โจ๐กโฒโฉ .
where ๐โจ๐ก,๐กโฒโฉ should be ๐ผโจ๐ก,๐กโฒโฉ
๐โจ๐ก,๐กโฒโฉ is then used when computing the attention ๐โจ๐ก,๐กโฒโฉ that ๐ฆโจ๐กโฉ should pay to ๐โจ๐กโฒโฉ .
where ๐โจ๐ก,๐กโฒโฉ should be ๐ผโจ๐ก,๐กโฒโฉ
Thanks for your report.