List your company

Question Answering on NewsQA (F1, EM)

80.5F1Oracle
9.790428.147746.50564.8623Apr 6, 2026Apr 17, 2026Apr 28, 2026May 9, 2026May 20, 2026May 31, 2026Jun 11, 2026

Evaluation results

MethodLinks
Oracle2026.0480.560.67
EvolveRouterPrompt Optimization=Enabled, Adaptive K=Enabled2026.0464.7441.67
EvolveRouter (w/o Adaptive K)Prompt Optimization=Enabled, Adaptive K=Disabled2026.0464.4842
EvolveRouter (w/o PO)Prompt Optimization=Disabled, Adaptive K=Enabled2026.0463.8141.33
AgentRouter2026.0463.0241.33
Best Agent2026.0462.837.11
GraphRouter2026.0462.6642.33
DOPAModel=Qwen3-1.7B2026.0461.4937.8
KNNModel=Qwen3-1.7B2026.0461.4537.3
MLP Router2026.0460.9437
Best LLM2026.0460.5539.8
CascadeRouting2026.0460.1739
AGAR2026.0660.1—
DOPAModel=LLaMA3.2-3B2026.0459.9733.2
RouteLLM2026.0459.9338
Majority Vote2026.0459.339.5
LLM-Blender2026.0459.337.33
KNNModel=LLaMA3.2-3B2026.0458.6132
AGARBackbone=GLM-4.1V-9B-Thinking2026.0658.5—
UniformBackbone=GLM-4.1V-9B-Thinking2026.0658.4—
VEA-ConBackbone=GLM-4.1V-9B-Thinking2026.0658.4—
Random Router2026.0458.1136.2
Plain VQABackbone=GLM-4.1V-9B-Thinking2026.0658.1—
VEA-BrBackbone=GLM-4.1V-9B-Thinking2026.0658—
Average2026.0457.9736.1
RandomBackbone=GLM-4.1V-9B-Thinking2026.0657.9—
VARBackbone=GLM-4.1V-9B-Thinking2026.0657.3—
KNN Router2026.0457.1534
Plain VQA2026.0657.1—
Uniform2026.0656.3—
Random2026.0655.3—
VAR2026.0655.3—
VEA-Br2026.0654.8—
VEA-Con2026.0654.5—
HybridLLM2026.0454.1931
SeCoBackbone=LLaMA-3.2-1B-Instruct, Compression constraint=16x2026.0550.8129.91
SeCoBackbone=LLaMA-3.2-1B-Instruct, Compression constraint=32x2026.0549.9129.91
CGRBackbone=GLM-4.1V-9B-Thinking2026.0649.7—
AGLA2026.0649.6—
AGLABackbone=GLM-4.1V-9B-Thinking2026.0649.5—
SeCoBackbone=Qwen3-4B-Instruct, Compression constraint=16x2026.0548.6327.4
CGR2026.0647—
SeCoBackbone=Qwen3-4B-Instruct, Compression constraint=32x2026.0546.5726.23
SAC (EPL)Backbone=Qwen3-4B-Instruct, Compression constraint=16x2026.0541.0820.25
500x (EPL)Backbone=Qwen3-4B-Instruct, Compression constraint=16x2026.0539.720.16
RAMBackbone=LLaMA-3.2-1B-Instruct, Compression constraint=16x2026.0535.1819.9
SAC (EPL)Backbone=Qwen3-4B-Instruct, Compression constraint=32x2026.0533.8815.86
500x (EPL)Backbone=Qwen3-4B-Instruct, Compression constraint=32x2026.0532.3114.84
ICAE (EPL)Backbone=Qwen3-4B-Instruct, Compression constraint=16x2026.0532.215.08
500x (DPL)Backbone=Qwen3-4B-Instruct, Compression constraint=16x2026.0531.0614.58
500x (DPL)Backbone=Qwen3-4B-Instruct, Compression constraint=32x2026.0530.514.34
RAMBackbone=LLaMA-3.2-1B-Instruct, Compression constraint=32x2026.0530.3317.57
SAC (EPL)Backbone=LLaMA-3.2-1B-Instruct, Compression constraint=16x2026.0527.9311.89
ICAE (EPL)Backbone=Qwen3-4B-Instruct, Compression constraint=32x2026.0527.8112.51
COMIBackbone=LLaMA-3.2-1B-Instruct, Compression constraint=16x2026.0527.1314.62
500x (EPL)Backbone=LLaMA-3.2-1B-Instruct, Compression constraint=16x2026.0526.5111.35
COMIBackbone=LLaMA-3.2-1B-Instruct, Compression constraint=32x2026.0524.7312.89
ICAE (EPL)Backbone=LLaMA-3.2-1B-Instruct, Compression constraint=16x2026.0524.3710.59
500x (DPL)Backbone=LLaMA-3.2-1B-Instruct, Compression constraint=16x2026.0523.0610.02
SAC (EPL)Backbone=LLaMA-3.2-1B-Instruct, Compression constraint=32x2026.0522.749.26
500x (EPL)Backbone=LLaMA-3.2-1B-Instruct, Compression constraint=32x2026.0522.219.76
500x (DPL)Backbone=LLaMA-3.2-1B-Instruct, Compression constraint=32x2026.0521.829.43
ICAE (EPL)Backbone=LLaMA-3.2-1B-Instruct, Compression constraint=32x2026.0520.658.86
RAMBackbone=Qwen3-4B-Instruct, Compression constraint=16x2026.0516.877.48
COMIBackbone=Qwen3-4B-Instruct, Compression constraint=16x2026.0516.316.77
RAMBackbone=Qwen3-4B-Instruct, Compression constraint=32x2026.0513.015.25
COMIBackbone=Qwen3-4B-Instruct, Compression constraint=32x2026.0512.514.77