List your company
ResearchFormerly WizWand

Mathematical Reasoning on GSM-Infinite (Avg)

17.1AccuracyXAttention
5.0368.16811.314.432Jul 29, 2025
Updated 6mo ago

Evaluation results

MethodLinks
XAttentionBackbone Model=Qwen2.5-7B-Inst2025.0717.1
MInferenceBackbone Model=Qwen2.5-7B-Inst2025.0717
TriangleMixBackbone Model=Qwen2.5-7B-Inst2025.0716.9
Ours + MInferenceBackbone Model=Qwen2.5-7B-Inst2025.0716.9
Ours + XAttentionBackbone Model=Qwen2.5-7B-Inst2025.0716.7
Qwen2.5-7B-InstBackbone Model=Qwen2.5-7B-Inst2025.0716.1
Ours + FlexPrefillBackbone Model=Qwen2.5-7B-Inst2025.0715.3
FlexPrefillBackbone Model=Qwen2.5-7B-Inst2025.0715.2
FlexPrefillBackbone Model=Llama-3.1-8B-Inst2025.0711.7
Ours + XAttentionBackbone Model=Llama-3.1-8B-Inst2025.078.8
XAttentionBackbone Model=Llama-3.1-8B-Inst2025.078.2
FlexPrefillBackbone Model=Llama-3-8B-Inst-262k2025.078.1
Llama-3.1-8B-InstBackbone Model=Llama-3.1-8B-Inst2025.077.6
TriangleMixBackbone Model=Llama-3.1-8B-Inst2025.077.4
MInferenceBackbone Model=Llama-3.1-8B-Inst2025.076.8
Ours + MInferenceBackbone Model=Llama-3.1-8B-Inst2025.076.8
Ours + FlexPrefillBackbone Model=Llama-3.1-8B-Inst2025.076.8
XAttentionBackbone Model=Llama-3-8B-Inst-262k2025.076.1
Llama-3-8B-Inst-262kBackbone Model=Llama-3-8B-Inst-262k2025.076
TriangleMixBackbone Model=Llama-3-8B-Inst-262k2025.076
Ours + XAttentionBackbone Model=Llama-3-8B-Inst-262k2025.075.8
MInferenceBackbone Model=Llama-3-8B-Inst-262k2025.075.7
Ours + MInferenceBackbone Model=Llama-3-8B-Inst-262k2025.075.5
Ours + FlexPrefillBackbone Model=Llama-3-8B-Inst-262k2025.075.5