Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

End-to-End Inference on LMSys-Chat-1M and ShareGPT traces

74.46p99 TTFT (ms)

Sem-MoE

72.64884.87997.11109.341Mar 6, 2025
Updated 5mo ago

Evaluation Results

MethodLinks
2025.03
74.46750.26
2025.03
75.63603.1
2025.03
80.96604.46
2025.03
83.87759.16
2025.03
84.87617.87
2025.03
85.72758.6
2025.03
87.24763.97
2025.03
88.69599.89
2025.03
92.17602.02
2025.03
98.1609.81
2025.03
100.74604.45
2025.03
101.86759.41
2025.03
103.69762.3
2025.03
104.76766.8
2025.03
107.36764.16
2025.03
107.73769.83
2025.03
111.19608.96
2025.03
119.76606.54