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

MoE LLMs

Benchmarks

Task NameDataset NameSOTA ResultTrend
Inference EfficiencyMoE LLMs DSV2-16B, QW3-30B, QW3-80B-I
Decode Speed (tokens/sec)12.46
9
Showing 1 of 1 rows