List your company

Science Question Answering on ARC Easy (Accuracy)

86.4AccuracyFull
26.18441.81757.4573.083Apr 17, 2025Jun 24, 2025Sep 1, 2025Nov 9, 2025Jan 16, 2026Mar 26, 2026Jun 3, 2026

Evaluation results

MethodLinks
FullModel=Qwen1.5MoE-A2.7B, E↓=100%, A↓=100%2026.0686.4
TENPModel=Qwen1.5MoE-A2.7B, E↓=60%, A↓=61.38%2026.0685.1
ENPModel=Qwen1.5MoE-A2.7B, E↓=60%, A↓=60%2026.0684.6
Gating ScoreModel=Qwen1.5MoE-A2.7B, E↓=60%, A↓=100%2026.0684.5
FullModel=DeepSeek-V2-Lite, E↓=100%, A↓=100%2026.0684.1
GEMMA-7BState=Dense2025.0681.65
Teacher (GPT-OSS-20B)Evaluation Mode=completion mode2026.0580.9
EASY-EPModel=Qwen1.5MoE-A2.7B, E↓=60%, A↓=100%2026.0680.1
FrequencyModel=Qwen1.5MoE-A2.7B, E↓=60%, A↓=100%2026.0680
TENPModel=DeepSeek-V2-Lite, E↓=60%, A↓=63.76%2026.0679.1
ENPModel=Qwen1.5MoE-A2.7B, E↓=30%, A↓=30%2026.0678.2
TENPModel=Qwen1.5MoE-A2.7B, E↓=30%, A↓=34.51%2026.0678.2
RandomModel=Qwen1.5MoE-A2.7B, E↓=60%, A↓=100%2026.0677.8
ENPModel=DeepSeek-V2-Lite, E↓=60%, A↓=60%2026.0677.8
EASY-EPModel=DeepSeek-V2-Lite, E↓=60%, A↓=100%2026.0676.7
LLAMA-2-7BState=Dense2025.0676.35
FrequencyModel=DeepSeek-V2-Lite, E↓=60%, A↓=100%2026.0675.3
DEEPSEEK-7BState=Dense2025.0675.25
VICUNA-1.3-7BState=Dense2025.0674.28
Gating ScoreModel=DeepSeek-V2-Lite, E↓=60%, A↓=100%2026.0674
CLIMBSize=950M2025.0473.57
TENPModel=DeepSeek-V2-Lite, E↓=30%, A↓=36.38%2026.0673.3
SmolLMSize=360M2025.0470.08
MASKLLMBackbone=LLAMA-2-7B2025.0669.57
MASKLLMBackbone=DEEPSEEK-7B2025.0668.1
CLIMBSize=350M2025.0467.38
MaskProBackbone=DEEPSEEK-7B2025.0666.74
SPARSEGPTBackbone=DEEPSEEK-7B2025.0666.37
MaskProBackbone=LLAMA-2-7B2025.0666.12
AMD-OLMoSize=1.2B2025.0465.95
GBLMBackbone=DEEPSEEK-7B2025.0665.82
MASKLLMBackbone=VICUNA-1.3-7B2025.0665.57
WANDABackbone=DEEPSEEK-7B2025.0665.49
Llama-3.2Size=1.2B2025.0465.49
PRUNER-ZEROBackbone=DEEPSEEK-7B2025.0664.94
Qwen2.5Size=490M2025.0464.6
SPARSEGPTBackbone=LLAMA-2-7B2025.0664.52
MaskProBackbone=VICUNA-1.3-7B2025.0664.23
GBLMBackbone=LLAMA-2-7B2025.0663.17
SPARSEGPTBackbone=VICUNA-1.3-7B2025.0662.92
GBLMBackbone=VICUNA-1.3-7B2025.0662.84
WANDABackbone=LLAMA-2-7B2025.0662.79
PRUNER-ZEROBackbone=VICUNA-1.3-7B2025.0662.33
MAGNITUDEBackbone=LLAMA-2-7B2025.0661.87
WANDABackbone=VICUNA-1.3-7B2025.0661.7
PRUNER-ZEROBackbone=LLAMA-2-7B2025.0661.32
X-GRAM-100%Backbone Scale=1.15B Scale, Cfg=4x, # Params=1.15B+14.94B, Evaluation Protocol=Zero/Few-shot2026.0461.1
X-GRAM-50%Backbone Scale=1.15B Scale, Cfg=4x, # Params=1.15B+7.47B, Evaluation Protocol=Zero/Few-shot2026.0460.9
X-GRAM-50%Backbone Scale=1.15B Scale, Cfg=2x, # Params=1.15B+3.73B, Evaluation Protocol=Zero/Few-shot2026.0460.5
TinyLlamaSize=1.1B2025.0460.31
ENPModel=DeepSeek-V2-Lite, E↓=30%, A↓=30%2026.0659.6
X-GRAM-100%Backbone Scale=1.15B Scale, Cfg=2x, # Params=1.15B+7.47B, Evaluation Protocol=Zero/Few-shot2026.0459.5
X-GRAM-100%Backbone Scale=0.73B Scale, Cfg=2x, # Params=0.73B+4.67B, Evaluation Protocol=Zero/Few-shot2026.0459.1
X-GRAM-100%Backbone Scale=0.73B Scale, Cfg=4x, # Params=0.73B+9.33B, Evaluation Protocol=Zero/Few-shot2026.0458.4
X-GRAM-50%Backbone Scale=0.73B Scale, Cfg=4x, # Params=0.73B+4.67B, Evaluation Protocol=Zero/Few-shot2026.0458.1
X-GRAM-100%Backbone Scale=1.15B Scale, Cfg=1x, # Params=1.15B+3.73B, Evaluation Protocol=Zero/Few-shot2026.0457.7
X-GRAM-50%Backbone Scale=0.73B Scale, Cfg=1x, # Params=0.73B+1.17B, Evaluation Protocol=Zero/Few-shot2026.0456.7
X-GRAM-50%Backbone Scale=1.15B Scale, Cfg=1x, # Params=1.15B+1.87B, Evaluation Protocol=Zero/Few-shot2026.0456.3
Gating ScoreModel=Qwen1.5MoE-A2.7B, E↓=30%, A↓=100%2026.0656.3
MoRTBackbone Scale=1.15B Scale, Cfg=4x, # Params=1.15B+14.94B, Evaluation Protocol=Zero/Few-shot2026.0455.6
EASY-EPModel=Qwen1.5MoE-A2.7B, E↓=30%, A↓=100%2026.0655.3
MoRTBackbone Scale=1.15B Scale, Cfg=2x, # Params=1.15B+7.47B, Evaluation Protocol=Zero/Few-shot2026.0455.1
MAGNITUDEBackbone=DEEPSEEK-7B2025.0654.92
MAGNITUDEBackbone=VICUNA-1.3-7B2025.0654.59
X-GRAM-100%Backbone Scale=0.73B Scale, Cfg=1x, # Params=0.73B+2.33B, Evaluation Protocol=Zero/Few-shot2026.0454
EngramBackbone Scale=0.73B Scale, Cfg=1x, # Params=0.73B+2.33B, Evaluation Protocol=Zero/Few-shot2026.0453.5
EngramBackbone Scale=0.73B Scale, Cfg=2x, # Params=0.73B+4.67B, Evaluation Protocol=Zero/Few-shot2026.0453.5
BaselineBackbone Scale=1.15B Scale, Cfg=N/A, # Params=1.15B, Evaluation Protocol=Zero/Few-shot2026.0453.2
X-GRAM-50%Backbone Scale=0.73B Scale, Cfg=2x, # Params=0.73B+2.33B, Evaluation Protocol=Zero/Few-shot2026.0453
FrequencyModel=Qwen1.5MoE-A2.7B, E↓=30%, A↓=100%2026.0652.6
RetokenBackbone Scale=1.15B Scale, Cfg=1x, # Params=1.15B+3.73B, Evaluation Protocol=Zero/Few-shot2026.0452.3
EngramBackbone Scale=1.15B Scale, Cfg=4x, # Params=1.15B+14.94B, Evaluation Protocol=Zero/Few-shot2026.0451.6
RetokenBackbone Scale=0.73B Scale, Cfg=2x, # Params=0.73B+4.67B, Evaluation Protocol=Zero/Few-shot2026.0451.2
RetokenBackbone Scale=1.15B Scale, Cfg=2x, # Params=1.15B+7.47B, Evaluation Protocol=Zero/Few-shot2026.0450.7
MoRTBackbone Scale=0.73B Scale, Cfg=4x, # Params=0.73B+9.33B, Evaluation Protocol=Zero/Few-shot2026.0449.6
MoRTBackbone Scale=1.15B Scale, Cfg=1x, # Params=1.15B+3.73B, Evaluation Protocol=Zero/Few-shot2026.0449.3
EngramBackbone Scale=1.15B Scale, Cfg=1x, # Params=1.15B+3.73B, Evaluation Protocol=Zero/Few-shot2026.0449.1
RetokenBackbone Scale=0.73B Scale, Cfg=4x, # Params=0.73B+9.33B, Evaluation Protocol=Zero/Few-shot2026.0448.8
EngramBackbone Scale=1.15B Scale, Cfg=2x, # Params=1.15B+7.47B, Evaluation Protocol=Zero/Few-shot2026.0448.6
MoRTBackbone Scale=0.73B Scale, Cfg=1x, # Params=0.73B+2.33B, Evaluation Protocol=Zero/Few-shot2026.0447.9
MoRTBackbone Scale=0.73B Scale, Cfg=2x, # Params=0.73B+4.67B, Evaluation Protocol=Zero/Few-shot2026.0447.9
RetokenBackbone Scale=1.15B Scale, Cfg=4x, # Params=1.15B+14.94B, Evaluation Protocol=Zero/Few-shot2026.0447.9
RetokenBackbone Scale=0.73B Scale, Cfg=1x, # Params=0.73B+2.33B, Evaluation Protocol=Zero/Few-shot2026.0446.1
EngramBackbone Scale=0.73B Scale, Cfg=4x, # Params=0.73B+9.33B, Evaluation Protocol=Zero/Few-shot2026.0443.5
BaselineBackbone Scale=0.73B Scale, Cfg=N/A, # Params=0.73B, Evaluation Protocol=Zero/Few-shot2026.0441.4
EASY-EPModel=DeepSeek-V2-Lite, E↓=30%, A↓=100%2026.0638.4
DO-ACPScoring=DO-ACP, K=42026.0536.7
CPScoring=CP, K=42026.0536.4
ACPScoring=ACP, K=42026.0535.6
RandomModel=DeepSeek-V2-Lite, E↓=60%, A↓=100%2026.0634.3
SFScoring=SF, K=42026.0534
ACPScoring=ACP, K=82026.0533.6
DO-ACPScoring=DO-ACP, K=82026.0533.5
SFScoring=SF, K=82026.0533.3
MaskProBackbone=GEMMA-7B2025.0632.92
WANDABackbone=GEMMA-7B2025.0632.66
GBLMBackbone=GEMMA-7B2025.0632.38
CPScoring=CP, K=82026.0532.2
SPARSEGPTBackbone=GEMMA-7B2025.0630.64
Random FFN + teacher attn2026.0528.5