List your company

Mathematical Reasoning on MinervaMath (Acc, Tokens, EFF)

54.04AccuracyThinkPrune(6k->3k)
19.678428.599237.5246.4408Oct 8, 2025Oct 31, 2025Nov 24, 2025Dec 18, 2025Jan 11, 2026Feb 4, 2026Feb 28, 2026

Evaluation results

MethodLinks
ThinkPrune(6k->3k)Prompt=step, Backbone=Qwen3-8B2026.0254.042,7251.98
D-SFT->D-RL3k->D-RL6kPrompt=step, Backbone=Qwen3-8B2026.0253.685,1801.03
D-SFT->D-RL6kPrompt=step, Backbone=Qwen3-8B2026.0253.685,0761.05
ThinkPrune(6k)Prompt=step, Backbone=Qwen3-8B2026.0253.683,8221.4
OriginalPrompt=step, Backbone=Qwen3-8B2026.0252.217,6400.68
D-SFT->D-RL6kPrompt=adaptive, Backbone=Qwen3-8B2026.02503,2091.55
D-SFT->D-RL6kPrompt=draft, Backbone=Qwen3-8B2026.0249.261,6772.93
D-SFT->D-RL3k->D-RL6kPrompt=adaptive, Backbone=Qwen3-8B2026.0248.532,9791.62
REO-RL(Q-Spec)*Prompt=step, Backbone=Qwen3-8B2026.0246.83,9521.18
D-SFT->D-RL3k->D-RL6kPrompt=draft, Backbone=Qwen3-8B2026.0246.329604.82
D-SFT->D-RL3kPrompt=draft, Backbone=Qwen3-8B2026.0244.491,1333.92
D-SFTPrompt=draft, Backbone=Qwen3-8B2026.0237.875496.88
Mean Acc.Backbone Model=Qwen3-8B2025.1034——
Self-Cert.Backbone Model=Qwen3-8B2025.1034——
Loc. non-uniBackbone Model=Qwen3-8B2025.1034——
Loc. uniBackbone Model=Qwen3-8B2025.1034——
Glob. uniBackbone Model=Qwen3-8B2025.1034——
High Conf.Backbone Model=Qwen3-8B2025.1033——
Low Ent.Backbone Model=Qwen3-8B2025.1033——
Glob. non-uniBackbone Model=Qwen3-8B2025.1033——
Loc. uniBackbone Model=DS-R1-Distill-Qwen-7B2025.1031——
Mean Acc.Backbone Model=DS-R1-Distill-Qwen-7B2025.1030——
Self-Cert.Backbone Model=DS-R1-Distill-Qwen-7B2025.1030——
High Conf.Backbone Model=DS-R1-Distill-Qwen-7B2025.1030——
Low Ent.Backbone Model=DS-R1-Distill-Qwen-7B2025.1030——
Glob. non-uniBackbone Model=DS-R1-Distill-Qwen-7B2025.1030——
Glob. uniBackbone Model=DS-R1-Distill-Qwen-7B2025.1028——
Loc. non-uniBackbone Model=DS-R1-Distill-Qwen-7B2025.1026——
Mean Acc.Backbone Model=DS-R1-Distill-Llama-8B2025.1023——
High Conf.Backbone Model=DS-R1-Distill-Llama-8B2025.1023——
Loc. non-uniBackbone Model=DS-R1-Distill-Llama-8B2025.1023——
Loc. uniBackbone Model=DS-R1-Distill-Llama-8B2025.1023——
Glob. uniBackbone Model=DS-R1-Distill-Llama-8B2025.1023——
Self-Cert.Backbone Model=DS-R1-Distill-Llama-8B2025.1022——
Low Ent.Backbone Model=DS-R1-Distill-Llama-8B2025.1022——
Glob. non-uniBackbone Model=DS-R1-Distill-Llama-8B2025.1021——