Reuse the same relative position bias in every layers #2258
Job | Run time |
---|---|
42m 11s | |
44m 9s | |
58m 39s | |
1h 48m 20s | |
14m 7s | |
7m 27s | |
6m 14s | |
10m 19s | |
9m 51s | |
32m 26s | |
6m 1s | |
9s | |
10m 26s | |
6m 6s | |
3m 38s | |
39s | |
0s | |
6h 0m 42s |
Job | Run time |
---|---|
42m 11s | |
44m 9s | |
58m 39s | |
1h 48m 20s | |
14m 7s | |
7m 27s | |
6m 14s | |
10m 19s | |
9m 51s | |
32m 26s | |
6m 1s | |
9s | |
10m 26s | |
6m 6s | |
3m 38s | |
39s | |
0s | |
6h 0m 42s |