Support multi-query attention for encoder-decoder models #2270
Job | Run time |
---|---|
34m 22s | |
8m 10s | |
7m 53s | |
12m 43s | |
10m 54s | |
6m 12s | |
41m 46s | |
1h 5m 54s | |
1h 8m 7s | |
1h 46m 2s | |
11m 32s | |
16s | |
40s | |
12m 50s | |
7m 45s | |
4m 21s | |
0s | |
6h 39m 27s |
Job | Run time |
---|---|
34m 22s | |
8m 10s | |
7m 53s | |
12m 43s | |
10m 54s | |
6m 12s | |
41m 46s | |
1h 5m 54s | |
1h 8m 7s | |
1h 46m 2s | |
11m 32s | |
16s | |
40s | |
12m 50s | |
7m 45s | |
4m 21s | |
0s | |
6h 39m 27s |