"vscode:/vscode.git/clone" did not exist on "459d822b5188dba051e21dfd15b6552543a4bbcf"
- 25 Mar, 2021 1 commit
-
-
TiagoMAntunes authored
-
- 23 Mar, 2021 3 commits
-
-
TiagoMAntunes authored
-
TiagoMAntunes authored
Bias now being calculated directly in MOELinear layer. Added corresponding CUDA changes. Updated forward and backward functions of MOELinear
-
TiagoMAntunes authored
-
- 13 Mar, 2021 3 commits
- 09 Mar, 2021 1 commit
-
-
Sengxian authored
-
- 08 Mar, 2021 1 commit
-
-
Rick Ho authored
-
- 06 Mar, 2021 1 commit
-
-
Jiezhong Qiu authored
-
- 05 Mar, 2021 1 commit
-
-
Rick Ho authored
-
- 04 Mar, 2021 1 commit
-
-
Rick Ho authored
-
- 02 Mar, 2021 1 commit
-
-
Rick Ho authored
-
- 01 Mar, 2021 4 commits
-
-
Jiezhong Qiu authored
-
Rick Ho authored
-
Rick Ho authored
-
Rick Ho authored
-
- 28 Feb, 2021 10 commits
-
-
Rick Ho authored
-
Rick Ho authored
-
Rick Ho authored
-
Rick Ho authored
-
Rick Ho authored
-
Rick Ho authored
-
Jiezhong Qiu authored
-
Jiezhong Qiu authored
-
Jiezhong Qiu authored
-
Jiezhong Qiu authored
remove author related info
-
- 26 Feb, 2021 9 commits
- 25 Feb, 2021 4 commits
-
-
Jiezhong Qiu authored
i.e., expand bias using torch.repeat_interleave directly
-
Jiezhong Qiu authored
-
Jiezhong Qiu authored
the true bias has been added in FMoeLinear
-
Jiezhong Qiu authored
move away layernorm/residual/dropout
-
