List your company

Multitask LLM Evaluation on MMLU, GSM8K, and HumanEval

88.33Average AccuracyVanilla MAS
57.722865.668973.61581.5611May 28, 2026

Evaluation results

MethodLinks
Vanilla MASBackbone=Qwen2.5-72B-Instruct, Topology=Random, Training-Free (TF)=true2026.0588.33171.940.51
AgentDropoutBackbone=Qwen2.5-72B-Instruct, Topology=FullConnected, Training-Free (TF)=false2026.0588.25155.810.57
Vanilla MASBackbone=Qwen2.5-72B-Instruct, Topology=Star, Training-Free (TF)=true2026.0587.96212.250.41
Vanilla MASBackbone=Qwen2.5-72B-Instruct, Topology=Layered, Training-Free (TF)=true2026.0587.62174.060.5
AgentDropoutBackbone=Qwen2.5-72B-Instruct, Topology=Star, Training-Free (TF)=false2026.0587.42155.330.56
Vanilla MASBackbone=Qwen2.5-72B-Instruct, Topology=Chain, Training-Free (TF)=true2026.0587.26192.680.45
AgentDropoutBackbone=Qwen2.5-72B-Instruct, Topology=Layered, Training-Free (TF)=false2026.0587.08137.840.63
SC-CoTBackbone=Qwen2.5-72B-Instruct, Topology=-, Training-Free (TF)=true2026.0586.9591.45—
Vanilla MASBackbone=Qwen2.5-72B-Instruct, Topology=FullConnected, Training-Free (TF)=true2026.0586.91196.870.44
AgentDropoutBackbone=Qwen2.5-72B-Instruct, Topology=Random, Training-Free (TF)=false2026.0586.88132.610.66
LLM-DebateBackbone=Qwen2.5-72B-Instruct, Topology=Debate, Training-Free (TF)=true2026.0586.47170.190.51
CONCATBackbone=Qwen2.5-14B-Instruct, Topology=Hybrid, Training-Free (TF)=true2026.0586.0230.172.85
CONCATBackbone=Qwen2.5-72B-Instruct, Topology=Hybrid, Training-Free (TF)=true2026.0585.7798.540.87
AgentDropoutBackbone=Qwen2.5-72B-Instruct, Topology=Chain, Training-Free (TF)=false2026.0585.45158.850.54
LLM-DebateBackbone=Qwen2.5-14B-Instruct, Topology=Debate, Training-Free (TF)=true2026.0585.3460.431.41
CoTBackbone=Qwen2.5-72B-Instruct, Topology=-, Training-Free (TF)=true2026.0585.2424.03—
Vanilla MASBackbone=Qwen2.5-14B-Instruct, Topology=Star, Training-Free (TF)=true2026.0585.1568.331.25
SC-CoTBackbone=Qwen2.5-14B-Instruct, Topology=-, Training-Free (TF)=true2026.0585.1130.42—
AgentDropoutBackbone=Qwen2.5-14B-Instruct, Topology=Star, Training-Free (TF)=false2026.0585.0548.041.77
Vanilla MASBackbone=Qwen2.5-14B-Instruct, Topology=Chain, Training-Free (TF)=true2026.0584.8957.561.47
AgentDropoutBackbone=Qwen2.5-14B-Instruct, Topology=FullConnected, Training-Free (TF)=false2026.0584.8848.691.74
Vanilla MASBackbone=Qwen2.5-14B-Instruct, Topology=Random, Training-Free (TF)=true2026.0584.8662.071.37
AgentDropoutBackbone=Qwen2.5-14B-Instruct, Topology=Layered, Training-Free (TF)=false2026.0584.7451.31.65
Vanilla MASBackbone=Qwen2.5-14B-Instruct, Topology=Layered, Training-Free (TF)=true2026.0584.6962.61.35
Vanilla MASBackbone=Qwen2.5-14B-Instruct, Topology=FullConnected, Training-Free (TF)=true2026.0584.4369.141.22
AgentDropoutBackbone=Qwen2.5-14B-Instruct, Topology=Chain, Training-Free (TF)=false2026.0584.2446.881.8
AgentDropoutBackbone=Qwen2.5-14B-Instruct, Topology=Random, Training-Free (TF)=false2026.0583.3646.851.78
CoTBackbone=Qwen2.5-14B-Instruct, Topology=-, Training-Free (TF)=true2026.0574.797.64—
AgentDropoutBackbone=Llama3-8B-Instruct, Topology=Chain, Training-Free (TF)=false2026.0565.7976.60.86
LLM-DebateBackbone=Llama3-8B-Instruct, Topology=Debate, Training-Free (TF)=true2026.0565.3793.780.7
CONCATBackbone=Llama3-8B-Instruct, Topology=Hybrid, Training-Free (TF)=true2026.0564.9741.731.56
AgentDropoutBackbone=Llama3-8B-Instruct, Topology=Random, Training-Free (TF)=false2026.0564.5669.980.92
Vanilla MASBackbone=Llama3-8B-Instruct, Topology=Layered, Training-Free (TF)=true2026.0564.12102.960.62
AgentDropoutBackbone=Llama3-8B-Instruct, Topology=Star, Training-Free (TF)=false2026.0563.1971.110.89
SC-CoTBackbone=Llama3-8B-Instruct, Topology=-, Training-Free (TF)=true2026.0562.725.11—
Vanilla MASBackbone=Llama3-8B-Instruct, Topology=FullConnected, Training-Free (TF)=true2026.0562.3103.870.6
Vanilla MASBackbone=Llama3-8B-Instruct, Topology=Chain, Training-Free (TF)=true2026.0562.1781.450.76
AgentDropoutBackbone=Llama3-8B-Instruct, Topology=FullConnected, Training-Free (TF)=false2026.0562.0392.280.67
Vanilla MASBackbone=Llama3-8B-Instruct, Topology=Star, Training-Free (TF)=true2026.0562105.430.59
AgentDropoutBackbone=Llama3-8B-Instruct, Topology=Layered, Training-Free (TF)=false2026.0561.2187.390.7
Vanilla MASBackbone=Llama3-8B-Instruct, Topology=Random, Training-Free (TF)=true2026.0561.1492.870.66
CoTBackbone=Llama3-8B-Instruct, Topology=-, Training-Free (TF)=true2026.0558.96.05—