Implement '--keep-split' to quantize model into several shards #10947
Job | Run time |
---|---|
2m 19s | |
50s | |
2m 1s | |
5m 47s | |
2m 6s | |
46s | |
2m 35s | |
49s | |
2m 52s | |
1m 45s | |
53s | |
53s | |
1m 41s | |
1m 57s | |
6m 0s | |
1m 59s | |
19m 40s | |
1m 31s | |
17m 1s | |
5m 42s | |
5m 42s | |
5m 9s | |
21m 21s | |
4m 19s | |
5m 37s | |
5m 37s | |
4m 46s | |
5m 12s | |
40s | |
3m 44s | |
17m 0s | |
0s | |
2h 38m 14s |