use recommended setting for use_reentrant w gradient checkpointing (#… #643
Job | Run time |
---|---|
12m 30s | |
12m 22s | |
12m 18s | |
10m 31s | |
5m 57s | |
4m 52s | |
4m 19s | |
7m 44s | |
1h 10m 33s |
Job | Run time |
---|---|
12m 30s | |
12m 22s | |
12m 18s | |
10m 31s | |
5m 57s | |
4m 52s | |
4m 19s | |
7m 44s | |
1h 10m 33s |