llama : improve BPE pre-processing + LLaMA 3 and Deepseek support #11441
Job | Run time |
---|---|
27m 24s | |
29m 30s | |
26m 3s | |
13m 40s | |
16m 53s | |
15m 34s | |
12m 7s | |
10m 30s | |
11m 11s | |
8m 3s | |
7m 29s | |
2h 58m 24s |
Job | Run time |
---|---|
27m 24s | |
29m 30s | |
26m 3s | |
13m 40s | |
16m 53s | |
15m 34s | |
12m 7s | |
10m 30s | |
11m 11s | |
8m 3s | |
7m 29s | |
2h 58m 24s |