fix: explicitly using mask kwarg to use MultiHeadDotProductAttention … #6
Job | Run time |
---|---|
4m 3s | |
12m 50s | |
3m 31s | |
2m 57s | |
2m 54s | |
3m 3s | |
11m 51s | |
3m 20s | |
4m 12s | |
4m 8s | |
2m 55s | |
3m 42s | |
59m 26s |
Job | Run time |
---|---|
4m 3s | |
12m 50s | |
3m 31s | |
2m 57s | |
2m 54s | |
3m 3s | |
11m 51s | |
3m 20s | |
4m 12s | |
4m 8s | |
2m 55s | |
3m 42s | |
59m 26s |