[WIP][BugFix]Fix accuracy issues caused by wrong etp_size passed into FusedMoEParallelConfig when using vLLM 0.9.0 #24
Job | Run time |
---|---|
5m 8s | |
3m 30s | |
4m 43s | |
3m 51s | |
4m 38s | |
3m 58s | |
25m 48s |
Job | Run time |
---|---|
5m 8s | |
3m 30s | |
4m 43s | |
3m 51s | |
4m 38s | |
3m 58s | |
25m 48s |