List your company
ResearchFormerly WizWand

Mathematical Reasoning on Minerva (Avg@k, Pass@k)

39.94Avg@kRPSFT
19.379224.717130.05535.3929May 8, 2026
Updated 5mo ago

Evaluation results

MethodLinks
RPSFTBackbone=Qwen2.5-7B-Instruct, RL Fine-tuning Stage=DAPO2026.0539.9459.19
SFTBackbone=Qwen2.5-7B-Instruct, RL Fine-tuning Stage=DAPO2026.0538.9758.09
IWBackbone=Qwen2.5-7B-Instruct, RL Fine-tuning Stage=DAPO2026.0536.7655.14
DFTBackbone=Qwen2.5-7B-Instruct, RL Fine-tuning Stage=DAPO2026.0535.9452.94
IWBackbone=Qwen2.5-7B-Instruct, RL Fine-tuning Stage=init2026.0534.0555.51
RPSFTBackbone=Qwen2.5-7B-Instruct, RL Fine-tuning Stage=init2026.0533.8256.62
DFTBackbone=Qwen2.5-7B-Instruct, RL Fine-tuning Stage=init2026.0533.7348.53
SFTBackbone=Qwen2.5-7B-Instruct, RL Fine-tuning Stage=init2026.0531.5258.09
SFTBackbone=Llama-3.1-8B-Instruct, RL Fine-tuning Stage=DAPO2026.0529.0951.47
IWBackbone=Llama-3.1-8B-Instruct, RL Fine-tuning Stage=DAPO2026.0528.7751.84
RPSFTBackbone=Llama-3.1-8B-Instruct, RL Fine-tuning Stage=DAPO2026.0528.5452.94
RPSFTBackbone=Llama-3.1-8B-Instruct, RL Fine-tuning Stage=init2026.0527.4850.74
SFTBackbone=Llama-3.1-8B-Instruct, RL Fine-tuning Stage=init2026.0526.8847.06
IWBackbone=Llama-3.1-8B-Instruct, RL Fine-tuning Stage=init2026.0526.7548.16
DFTBackbone=Qwen2.5-3B-Instruct, RL Fine-tuning Stage=init2026.0526.5642.28
RPSFTBackbone=Qwen2.5-3B-Instruct, RL Fine-tuning Stage=DAPO2026.0526.4750.37
SFTBackbone=Qwen2.5-3B-Instruct, RL Fine-tuning Stage=DAPO2026.0525.3247.79
IWBackbone=Qwen2.5-3B-Instruct, RL Fine-tuning Stage=DAPO2026.0524.5445.96
RPSFTBackbone=Qwen2.5-3B-Instruct, RL Fine-tuning Stage=init2026.0523.8547.06
IWBackbone=Qwen2.5-3B-Instruct, RL Fine-tuning Stage=init2026.0523.7145.22
DFTBackbone=Qwen2.5-3B-Instruct, RL Fine-tuning Stage=DAPO2026.0523.6739.71
DFTBackbone=Llama-3.1-8B-Instruct, RL Fine-tuning Stage=DAPO2026.0522.9838.6
SFTBackbone=Qwen2.5-3B-Instruct, RL Fine-tuning Stage=init2026.0522.7547.42
DFTBackbone=Llama-3.1-8B-Instruct, RL Fine-tuning Stage=init2026.0520.1737.5