Dynamic memory allocation. Drop Baichuan/InternLM support in favor of llama.cpp. #251
Job | Run time |
---|---|
3m 59s | |
2m 56s | |
2m 48s | |
2m 48s | |
3m 1s | |
2m 37s | |
2m 17s | |
2m 2s | |
2m 42s | |
25m 10s |
Job | Run time |
---|---|
3m 59s | |
2m 56s | |
2m 48s | |
2m 48s | |
3m 1s | |
2m 37s | |
2m 17s | |
2m 2s | |
2m 42s | |
25m 10s |