List your company

Common Sense Reasoning on ARC Challenge

89.6AccuracyDASD
19.60837.77955.9574.121Nov 6, 2025Dec 13, 2025Jan 19, 2026Feb 25, 2026Apr 3, 2026May 10, 2026Jun 17, 2026

Evaluation results

MethodLinks
DASDBase Model=Qwen3-4B2026.0589.674.3
Hint-decodingBase Model=Qwen3-4B2026.0588.971.6
Self-DistillationBase Model=Qwen3-4B2026.0588.869.7
Mask-PPLBase Model=Qwen3-4B2026.0588.471.1
OriginBase Model=Qwen3-4B2026.058870.8
DASDBase Model=Llama3.2-3B2026.057856.7
Hint-decodingBase Model=Llama3.2-3B2026.057756.3
Mask-PPLBase Model=Llama3.2-3B2026.0576.254.7
DASDBase Model=Gemma2-2B2026.0575.547
Mask-PPLBase Model=Gemma2-2B2026.0574.945.4
Hint-decodingBase Model=Gemma2-2B2026.0574.545.9
Self-DistillationBase Model=Llama3.2-3B2026.0574.453.1
OriginBase Model=Llama3.2-3B2026.057452.3
Self-DistillationBase Model=Gemma2-2B2026.0573.344.8
OriginBase Model=Gemma2-2B2026.057043
VanillaBackbone=Mixtral-8x7B-v0.1, Sparsity=0%2025.1156.7—
PuzzleMoEBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.1155.3—
PuzzleMoEBackbone=Mixtral-8x7B-v0.1, Sparsity=50%2025.1153.8—
STUNBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.1152.7—
VanillaBackbone=Qwen3-MoE-30B-A3B, Sparsity=0%2025.1152.7—
NAEEBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.1152—
PuzzleMoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=25%2025.1151.6—
D2Backbone=Mixtral-8x7B-v0.1, Sparsity=20%2025.1151—
PuzzleMoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=50%2025.1151—
HC-SMoEBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.1150.3—
Sub-MoEBackbone=Mixtral-8x7B-v0.1, Sparsity=25%2025.1149—
WandaBackbone=Mixtral-8x7B-v0.1, Sparsity=2:42025.1148.3—
WandaBackbone=Qwen3-MoE-30B-A3B, Sparsity=2:42025.1148.2—
NAEEBackbone=Mixtral-8x7B-v0.1, Sparsity=50%2025.1148.1—
D2Backbone=Mixtral-8x7B-v0.1, Sparsity=40%2025.1147—
Sub-MoEBackbone=Mixtral-8x7B-v0.1, Sparsity=50%2025.1145—
VanillaBackbone=Deepseek-MoE-16b, Sparsity=0%2025.1144.6—
PuzzleMoEBackbone=Deepseek-MoE-16b, Sparsity=25%2025.1144—
Sub-MoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=25%2025.1144—
PuzzleMoEBackbone=Deepseek-MoE-16b, Sparsity=50%2025.1143—
D2Backbone=Qwen3-MoE-30B-A3B, Sparsity=20%2025.1142.1—
HC-SMoEBackbone=Mixtral-8x7B-v0.1, Sparsity=50%2025.1141.1—
D2Backbone=Deepseek-MoE-16b, Sparsity=20%2025.1141—
VanillaBackbone=Qwen1.5-MoE-A2.7B, Sparsity=0%2025.1141—
PuzzleMoEBackbone=Qwen1.5-MoE-A2.7B, Sparsity=25%2025.1140.9—
PuzzleMoEBackbone=Qwen1.5-MoE-A2.7B, Sparsity=50%2025.1140.7—
HC-SMoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=25%2025.1140.1—
Sub-MoEBackbone=Deepseek-MoE-16b, Sparsity=25%2025.1140—
Sub-MoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=50%2025.1140—
WandaBackbone=Qwen1.5-MoE-A2.7B, Sparsity=2:42025.1138.7—
WandaBackbone=Deepseek-MoE-16b, Sparsity=2:42025.1137.4—
HC-SMoEBackbone=Deepseek-MoE-16b, Sparsity=25%2025.1136.7—
D2Backbone=Deepseek-MoE-16b, Sparsity=40%2025.1136—
D2Backbone=Qwen3-MoE-30B-A3B, Sparsity=40%2025.1135.7—
HC-SMoEBackbone=Qwen1.5-MoE-A2.7B, Sparsity=25%2025.1134.8—
HC-SMoEBackbone=Qwen3-MoE-30B-A3B, Sparsity=50%2025.1134.3—
D2Backbone=Qwen1.5-MoE-A2.7B, Sparsity=20%2025.1133.6—
Sub-MoEBackbone=Deepseek-MoE-16b, Sparsity=50%2025.1132—
D2Backbone=Qwen1.5-MoE-A2.7B, Sparsity=40%2025.1127.8—
RegMix-DProxy models=512, Mode=Online2026.0627.57—
RegMix-DProxy models=128, Mode=Offline2026.0627.23—
RegMix-DProxy models=128, Mode=Online2026.0626.55—
RegMix2026.0626.39—
RegMix-DProxy models=512, Mode=Offline2026.0626.32—
DoReMi2026.0625.78—
Human2026.0625.61—
HC-SMoEBackbone=Qwen1.5-MoE-A2.7B, Sparsity=50%2025.1123.9—
HC-SMoEBackbone=Deepseek-MoE-16b, Sparsity=50%2025.1122.3—