List your company
ResearchFormerly WizWand

Mathematical Reasoning on Math Reasoning Suite Average

63Average AccuracyReasoning Palette (Linear Decay)
8.71222.80636.950.994Dec 19, 2025Jan 14, 2026Feb 9, 2026Mar 8, 2026Apr 3, 2026Apr 29, 2026May 26, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Reasoning Palette (Linear Decay)Backbone=Qwen3-8B-Base, RL Algorithm=RLOO2025.1263
Reasoning Palette (Two-Phase)Backbone=Qwen3-8B-Base, RL Algorithm=RLOO2025.1262.28
Reasoning Palette (Linear Decay)Backbone=Qwen3-4B-Base, RL Algorithm=RLOO2025.1261.28
Reasoning Palette (Linear Decay)Backbone=Qwen3-8B-Base, RL Algorithm=GRPO2025.1261.17
Reasoning Palette (Two-Phase)Backbone=Qwen3-4B-Base, RL Algorithm=GRPO2025.1260.77
Reasoning Palette (Two-Phase)Backbone=Qwen3-4B-Base, RL Algorithm=RLOO2025.1260.69
Reasoning Palette (Two-Phase)Backbone=Qwen3-8B-Base, RL Algorithm=GRPO2025.1260.57
Reasoning Palette (Linear Decay)Backbone=Qwen3-4B-Base, RL Algorithm=GRPO2025.1260.12
GRPOBackbone=Qwen3-8B-Base, RL Algorithm=GRPO2025.1260.05
RLOOBackbone=Qwen3-8B-Base, RL Algorithm=RLOO2025.1259.91
RLOOBackbone=Qwen3-4B-Base, RL Algorithm=RLOO2025.1259.39
GRPOBackbone=Qwen3-4B-Base, RL Algorithm=GRPO2025.1258.27
Qwen2.5-7B-Math + SFT + RL-TCERBase Model=Qwen2.5-7B-Math, Training Strategy=SFT + RL-TCER2026.0450.2
Qwen2.5-7B-Math + SFT + RL-EndoRBase Model=Qwen2.5-7B-Math, Training Strategy=SFT + RL-EndoR2026.0448.9
ExOPDDistillation Setting=Single-Teacher Distillation2026.0248
ExOPDDistillation Setting=Multi-Teacher Distillation2026.0247.7
OPDDistillation Setting=Single-Teacher Distillation2026.0246.5
OPDDistillation Setting=Multi-Teacher Distillation2026.0246.4
TeacherDistillation Setting=Baseline2026.0246
ExPODistillation Setting=Single-Teacher Distillation2026.0245.8
ExPODistillation Setting=Multi-Teacher Distillation2026.0245
SFTDistillation Setting=Multi-Teacher Distillation2026.0244.3
Qwen2.5-7B-Math + SFTBase Model=Qwen2.5-7B-Math, Training Strategy=SFT2026.0444.1
Llama3.1-8B-Instruct + SFT + RL-TCERBase Model=Llama3.1-8B-Instruct, Training Strategy=SFT + RL-TCER2026.0433.6
Llama3.1-8B-Instruct + SFT + RL-EndoRBase Model=Llama3.1-8B-Instruct, Training Strategy=SFT + RL-EndoR2026.0431.9
Llama3.1-8B-Instruct + SFTBase Model=Llama3.1-8B-Instruct, Training Strategy=SFT2026.0430.9
DAPO++Backbone=Qwen3-8B, Sampling Strategy=Mixed2026.0529.6
DAPO-Math-17kBackbone=Qwen3-8B, Sampling Strategy=Mixed2026.0529.3
DeepScaleRBackbone=Qwen3-8B, Sampling Strategy=Mixed2026.0526.1
DeepMath-103KBackbone=Qwen3-8B, Sampling Strategy=Mixed2026.0525.1
Skywork-OR1-RL-DataBackbone=Qwen3-8B, Sampling Strategy=Mixed2026.0525.1
OpenR1-Math-220kBackbone=Qwen3-8B, Sampling Strategy=Mixed2026.0525
Reasoning Palette (Linear Decay)Backbone=Qwen3-1.7B-Base, RL Algorithm=GRPO2025.1224.05
Reasoning Palette (Linear Decay)Backbone=Qwen3-1.7B-Base, RL Algorithm=RLOO2025.1223.49
Reasoning Palette (Two-Phase)Backbone=Qwen3-1.7B-Base, RL Algorithm=GRPO2025.1222.91
Reasoning Palette (Two-Phase)Backbone=Qwen3-1.7B-Base, RL Algorithm=RLOO2025.1222.8
RLOOBackbone=Qwen3-1.7B-Base, RL Algorithm=RLOO2025.1221.77
GRPOBackbone=Qwen3-1.7B-Base, RL Algorithm=GRPO2025.1221.18
Qwen2.5-7B-MathBase Model=Qwen2.5-7B-Math, Training Strategy=None2026.0419.1
Qwen3-8B-BaseBackbone=Qwen3-8B, Sampling Strategy=Mixed2026.0518.6
Llama3.1-8B-InstructBase Model=Llama3.1-8B-Instruct, Training Strategy=None2026.0417.6
DAPO++Backbone=Qwen3-1.7B, Sampling Strategy=Mixed2026.0515.7
StudentDistillation Setting=Baseline2026.0215.4
DeepMath-103KBackbone=Qwen3-1.7B, Sampling Strategy=Mixed2026.0515.4
Skywork-OR1-RL-DataBackbone=Qwen3-1.7B, Sampling Strategy=Mixed2026.0515.1
DAPO-Math-17kBackbone=Qwen3-1.7B, Sampling Strategy=Mixed2026.0515
DeepScaleRBackbone=Qwen3-1.7B, Sampling Strategy=Mixed2026.0514.7
OpenR1-Math-220kBackbone=Qwen3-1.7B, Sampling Strategy=Mixed2026.0514
Qwen3-1.7B-BaseBackbone=Qwen3-1.7B, Sampling Strategy=Mixed2026.0510.8