List your company
ResearchFormerly WizWand

Mathematical Reasoning on GSM-Infinite (8K)

22.9AccuracyTriangleMix
2.627.88513.1518.415Jul 29, 2025
Updated 6mo ago

Evaluation results

MethodLinks
TriangleMixBackbone Model=Qwen2.5-7B-Inst2025.0722.9
Ours + XAttentionBackbone Model=Qwen2.5-7B-Inst2025.0721.6
Ours + MInferenceBackbone Model=Qwen2.5-7B-Inst2025.0721.3
XAttentionBackbone Model=Qwen2.5-7B-Inst2025.0721.3
Qwen2.5-7B-InstBackbone Model=Qwen2.5-7B-Inst2025.0721
MInferenceBackbone Model=Qwen2.5-7B-Inst2025.0720.4
Ours + FlexPrefillBackbone Model=Qwen2.5-7B-Inst2025.0718.6
FlexPrefillBackbone Model=Qwen2.5-7B-Inst2025.0717.7
FlexPrefillBackbone Model=Llama-3.1-8B-Inst2025.078.6
Ours + XAttentionBackbone Model=Llama-3.1-8B-Inst2025.077.6
XAttentionBackbone Model=Llama-3.1-8B-Inst2025.077.3
TriangleMixBackbone Model=Llama-3.1-8B-Inst2025.077
Llama-3.1-8B-InstBackbone Model=Llama-3.1-8B-Inst2025.076.9
MInferenceBackbone Model=Llama-3.1-8B-Inst2025.076.5
Ours + MInferenceBackbone Model=Llama-3.1-8B-Inst2025.076.3
Ours + FlexPrefillBackbone Model=Llama-3.1-8B-Inst2025.076.3
FlexPrefillBackbone Model=Llama-3-8B-Inst-262k2025.075.3
Ours + MInferenceBackbone Model=Llama-3-8B-Inst-262k2025.073.8
Ours + FlexPrefillBackbone Model=Llama-3-8B-Inst-262k2025.073.8
Llama-3-8B-Inst-262kBackbone Model=Llama-3-8B-Inst-262k2025.073.7
XAttentionBackbone Model=Llama-3-8B-Inst-262k2025.073.5
TriangleMixBackbone Model=Llama-3-8B-Inst-262k2025.073.4
MInferenceBackbone Model=Llama-3-8B-Inst-262k2025.073.4
Ours + XAttentionBackbone Model=Llama-3-8B-Inst-262k2025.073.4