List your company
ResearchFormerly WizWand

Mathematical Reasoning on AMC (Accuracy %)

95.18Accuracy (%)Qwen3-30B A3B-Instruct-2507
60.672869.631478.5987.5486May 21, 2025Jul 23, 2025Sep 24, 2025Nov 26, 2025Jan 28, 2026Apr 1, 2026Jun 3, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Qwen3-30B A3B-Instruct-2507Orchestration Strategy=Majority Vote, Sampling Protocol (mean@8)=true2026.0495.18
ParaManagerBackbone=ParaManager-4B, Orchestration Strategy=Unified, Sampling Protocol (mean@8)=true2026.0495.18
Meta Agent SearchOrchestration Strategy=Static Workflow, Sampling Protocol (mean@8)=true2026.0494.58
GPT-OSS-20BOrchestration Strategy=Base+Tool, Sampling Protocol (mean@8)=true2026.0493.98
PuppeteerOrchestration Strategy=Serial Orchestration, Sampling Protocol (mean@8)=true2026.0493.98
EvoFlowOrchestration Strategy=Static Workflow, Sampling Protocol (mean@8)=true2026.0492.47
Qwen3-30B A3B-Instruct-2507Orchestration Strategy=Self Refine, Sampling Protocol (mean@8)=true2026.0492.17
GPT-OSS-20BOrchestration Strategy=Majority Vote, Sampling Protocol (mean@8)=true2026.0492.17
GPT-OSS-20BOrchestration Strategy=Self Refine, Sampling Protocol (mean@8)=true2026.0492.17
ParaManager-MonoBackbone=ParaManager-4B, Orchestration Strategy=Mono, Sampling Protocol (mean@8)=true2026.0492.17
ParaManager-SFTBackbone=ParaManager-4B, Orchestration Strategy=SFT, Sampling Protocol (mean@8)=true2026.0492.02
Qwen3-30B A3B-Instruct-2507Orchestration Strategy=Base, Sampling Protocol (mean@8)=true2026.0491.87
RouterOrchestration Strategy=Static Workflow, Sampling Protocol (mean@8)=true2026.0491.87
ParaManager-SerialBackbone=ParaManager-4B, Orchestration Strategy=Serial, Sampling Protocol (mean@8)=true2026.0491.87
Qwen3-4B Instruct-2507Orchestration Strategy=Majority Vote, Sampling Protocol (mean@8)=true2026.0491.57
GPT-OSS-20BOrchestration Strategy=Base, Sampling Protocol (mean@8)=true2026.0490.66
Qwen3-30B A3B-Instruct-2507Orchestration Strategy=Base+Tool, Sampling Protocol (mean@8)=true2026.0490.36
ToolOrchestraOrchestration Strategy=Serial Orchestration, Sampling Protocol (mean@8)=true2026.0490.36
HAPOBackbone=Qwen3-14B2025.0989.66
Qwen3-4B Instruct-2507Orchestration Strategy=Self Refine, Sampling Protocol (mean@8)=true2026.0489.61
Qwen3-4B Instruct-2507Orchestration Strategy=Base+Tool, Sampling Protocol (mean@8)=true2026.0488.25
DAPO w/ Forking TokensBackbone=Qwen3-14B2025.0987.31
EDGE-GRPOBackbone=Qwen3-14B2025.0987.15
Entropy EnvBackbone=Qwen3-14B2025.0987.04
Qwen3-4B Instruct-2507Orchestration Strategy=Base, Sampling Protocol (mean@8)=true2026.0486.9
ArcherBackbone=Qwen3-14B2025.0986.73
Vanilla DAPOBackbone=Qwen3-14B2025.0985.52
HAPOBackbone=Qwen2.5-Math-7B2025.0985.47
Vanilla GRPOBackbone=Qwen3-14B2025.0983.91
EDGE-GRPOBackbone=Qwen2.5-Math-7B2025.0983.23
Entropy EnvBackbone=Qwen2.5-Math-7B2025.0982.94
ArcherBackbone=Qwen2.5-Math-7B2025.0982.53
Vanilla DAPOBackbone=Qwen2.5-Math-7B2025.0981.99
HAPOBackbone=Qwen3-8B2025.0981.77
DAPO w/ Forking TokensBackbone=Qwen2.5-Math-7B2025.0980.55
EDGE-GRPOBackbone=Qwen3-8B2025.0980.47
DAPO w/ Forking TokensBackbone=Qwen3-8B2025.0980.26
Vanilla GRPOBackbone=Qwen2.5-Math-7B2025.0980.14
Entropy EnvBackbone=Qwen3-8B2025.0980.05
ArcherBackbone=Qwen3-8B2025.0978.47
Vanilla DAPOBackbone=Qwen3-8B2025.0978.37
TAPOBase Model=Qwen2.5-Math-7B2025.0777.5
TemplateRLBackbone=Qwen2.5-Math-7B-Base, RL Baseline=GRPO2025.0577.5
Vanilla GRPOBackbone=Qwen3-8B2025.0976.84
HAPOBackbone=Qwen2.5-Math-1.5B2025.0975.69
AISModel=Qwen3.5-9B, Bitwidth=FP82026.0575.5
DAPO w/ Forking TokensBackbone=Qwen2.5-Math-1.5B2025.0974.8
EDGE-GRPOBackbone=Qwen2.5-Math-1.5B2025.0974.48
RL (BF16)Model=Qwen3.5-9B, Bitwidth=BF162026.0574.2
Entropy EnvBackbone=Qwen2.5-Math-1.5B2025.0972.88
Vanilla DAPOBackbone=Qwen2.5-Math-1.5B2025.0972.85
ArcherBackbone=Qwen2.5-Math-1.5B2025.0972.17
Vanilla GRPOBackbone=Qwen2.5-Math-1.5B2025.0971.23
SPICEBackbone=Qwen3-8B-Base, #Data=20K2026.0570
AISModel=Qwen3-8B, Bitwidth=FP82026.0569.75
DDRLBase Model=Qwen2.5-Math-7B2026.0469
RL (BF16)Model=Qwen3-8B, Bitwidth=BF162026.0568.5
RL-PLUSBase Model=Qwen2.5-Math-7B, Training Method=RL-PLUS2025.0768.1
RL-PLUSBase Model=Qwen2.5-Math-7B2025.0768.1
TTRLBase Model=Qwen2.5-Math-7B2026.0468.1
FlashRL (TIS)Model=Qwen3.5-9B, Bitwidth=FP82026.0567.5
TemplateRLBackbone=Qwen2.5-Math-7B-Base, RL Baseline=RLOO2025.0567.5
TEPOBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0466.41
Hi-CoTModel=Qwen3-4B-Instruct-25072026.0366.3
DAPOBase Model=Qwen2.5-Math-7B2025.0766.3
GRPO/DAPOBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0466.1
DoTS (ExGRPO + ReLIFT)Backbone=Qwen2.5-Math-7B, Integration=Decoupled Test-time Synthesis2026.0566.1
RL (FP8 Rollout)Model=Qwen3.5-9B, Bitwidth=FP82026.0565.8
ExGRPOBackbone=Qwen2.5-Math-7B2026.0565.7
LUFFYBase Model=Qwen2.5-Math-7B2025.0765.6
CoTModel=Qwen3-4B-Instruct-25072026.0365.1
Plan-and-SolveModel=Qwen3-4B-Instruct-25072026.0365.1
PivotTraceBackbone=Deepseek-R1-Distill-Qwen-1.5B2026.0665.1
QueSTBackbone=Qwen3-8B-Base2026.0565
CoEBackbone=Deepseek-R1-Distill-Qwen-1.5B2026.0665
CoT-KineticsBackbone=Deepseek-R1-Distill-Qwen-1.5B2026.0664.8
EntropyBackbone=Deepseek-R1-Distill-Qwen-1.5B2026.0664.8
D2EvoBackbone=Qwen3-8B-Base, #Data=0.3K, Iteration=Iter 22026.0564.76
D2EvoBackbone=Qwen3-8B-Base, #Data=0.4K, Iteration=Iter 32026.0564.76
LUFFYBackbone=Qwen2.5-Math-7B2026.0564.6
ReLIFTBackbone=Qwen2.5-Math-7B2026.0564.4
D2EvoBackbone=Qwen3-4B-Base, #Data=0.1K, Iteration=Iter 32026.0564.38
D2EvoBackbone=Qwen3-8B-Base, #Data=1K, Iteration=Iter 12026.0564.34
ReLIFTBase Model=Qwen2.5-Math-7B2025.0764.3
FlashRL (TIS)Model=Qwen3-8B, Bitwidth=FP82026.0563.4
ConsistencyBackbone=Deepseek-R1-Distill-Qwen-1.5B, Multiple stochastic inferences=true2026.0663.4
DoTS (SFT + On-Policy RL)Backbone=Qwen2.5-Math-7B, Integration=Decoupled Test-time Synthesis2026.0563.2
Full DataBackbone=Qwen3-8B-Base, #Data=19K2026.0563.17
StandardModel=Qwen3-4B-Instruct-25072026.0362.7
SFT+GRPOBase Model=Qwen2.5-Math-7B2025.0762.7
SFT+RLBackbone=Qwen2.5-Math-7B, Protocol=Sequential training2026.0562.7
Self-CertaintyBackbone=Deepseek-R1-Distill-Qwen-1.5B2026.0662.6
TLMBackbone=Qwen3-8B-Base2026.0562.5
AZRBackbone=Qwen3-8B-Base2026.0562.5
Full DataBackbone=Qwen3-4B-Base, #Data=19K2026.0562.42
TIES-Merging* (SFT + On-Policy RL)Backbone=Qwen2.5-Math-7B, Integration=Model Merging with DoTS sparsification2026.0562.3
GRPOBackbone=Qwen2.5-Math-7B2026.0462
GRPOBase Model=Qwen2.5-Math-7B, Training Method=GRPO2025.0762
GRPOBase Model=Qwen2.5-Math-7B2025.0762
On-Policy RLBackbone=Qwen2.5-Math-7B2026.0562