List your company
ResearchFormerly WizWand

Mathematical Reasoning on MATH (val)

41.6AccuracyEVPO
0.93611.49322.0532.607May 30, 2024Sep 26, 2024Jan 23, 2025May 22, 2025Sep 18, 2025Jan 15, 2026May 15, 2026
Updated 5mo ago

Evaluation results

MethodLinks
EVPOBackbone=Qwen2.5-7B-Instruct2026.0441.6
DAPOBackbone=Qwen2.5-7B-Instruct, Method Category=Critic-free2026.0438.5
GRPOBackbone=Qwen2.5-7B-Instruct, Method Category=Critic-free2026.0437.9
PPOBackbone=Qwen2.5-7B-Instruct, Method Category=Critic-based2026.0437.3
StarPO-SBackbone=Qwen2.5-7B-Instruct, Method Category=Critic-based2026.0437.1
Base LLMBackbone=Qwen2.5-7B-Instruct, Method Category=None2026.0433.2
TAIATraining Dataset=Alpaca-GPT4, Model=LLaMA3-8B, FT Method=LoRA, Infer Mode=TAIA2024.0528.34
TAIATraining Dataset=Alpaca-GPT4, Model=LLaMA3-8B, FT Method=MOLORA, Infer Mode=TAIA2024.0528.16
LLaMA3-8BTraining Dataset=Base Model, Model=LLaMA3-8B2024.0527.92
TAIATraining Dataset=CoT-Collection, Model=LLaMA3-8B, FT Method=LoRA, Infer Mode=TAIA2024.0526.28
TAIATraining Dataset=CoT-Collection, Model=LLaMA3-8B, FT Method=MOLORA, Infer Mode=TAIA2024.0525.46
MOLORATraining Dataset=Alpaca-GPT4, Model=LLaMA3-8B, FT Method=MOLORA, Infer Mode=Vanilla2024.0525.38
LoRATraining Dataset=Alpaca-GPT4, Model=LLaMA3-8B, FT Method=LoRA, Infer Mode=Vanilla2024.0525.26
TAIATraining Dataset=Alpaca-GPT4, Model=Qwen1.5-7B, FT Method=MOLORA, Infer Mode=TAIA2024.0525.04
TAIATraining Dataset=Alpaca-GPT4, Model=Qwen1.5-7B, FT Method=LoRA, Infer Mode=TAIA2024.0524.98
Qwen1.5-7BTraining Dataset=Base Model, Model=Qwen1.5-7B2024.0520.3
TAIATraining Dataset=CoT-Collection, Model=Qwen1.5-7B, FT Method=LoRA, Infer Mode=TAIA2024.0520.28
TAIATraining Dataset=CoT-Collection, Model=Qwen1.5-7B, FT Method=MOLORA, Infer Mode=TAIA2024.0519.74
MOLORATraining Dataset=Alpaca-GPT4, Model=Qwen1.5-7B, FT Method=MOLORA, Infer Mode=Vanilla2024.0518.12
LoRATraining Dataset=Alpaca-GPT4, Model=Qwen1.5-7B, FT Method=LoRA, Infer Mode=Vanilla2024.0517.9
MOLORATraining Dataset=CoT-Collection, Model=LLaMA3-8B, FT Method=MOLORA, Infer Mode=Vanilla2024.0517.7
LoRATraining Dataset=CoT-Collection, Model=LLaMA3-8B, FT Method=LoRA, Infer Mode=Vanilla2024.0516.12
MOLORATraining Dataset=CoT-Collection, Model=Qwen1.5-7B, FT Method=MOLORA, Infer Mode=Vanilla2024.0513.38
LoRATraining Dataset=CoT-Collection, Model=Qwen1.5-7B, FT Method=LoRA, Infer Mode=Vanilla2024.0513.22
TAIATraining Dataset=Alpaca-GPT4, Model=Qwen1.5-1.8B, FT Method=LoRA, Infer Mode=TAIA2024.0510.82
TAIATraining Dataset=Alpaca-GPT4, Model=Qwen1.5-1.8B, FT Method=MOLORA, Infer Mode=TAIA2024.0510.2
TAIATraining Dataset=CoT-Collection, Model=Qwen1.5-1.8B, FT Method=LoRA, Infer Mode=TAIA2024.059.64
BA-LoRAPEFT Method=BA-LoRA, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.089.47
TAIATraining Dataset=CoT-Collection, Model=Qwen1.5-1.8B, FT Method=MOLORA, Infer Mode=TAIA2024.059.08
CorDA++PEFT Method=CorDA++, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.088.95
TAIATraining Dataset=Alpaca-GPT4, Model=LLaMA2-7B, FT Method=MOLORA, Infer Mode=TAIA2024.058.82
CorDAPEFT Method=CorDA, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.088.52
MOLORATraining Dataset=Alpaca-GPT4, Model=Qwen1.5-1.8B, FT Method=MOLORA, Infer Mode=Vanilla2024.058.44
TAIATraining Dataset=CoT-Collection, Model=LLaMA2-7B, FT Method=LoRA, Infer Mode=TAIA2024.058.44
LoCOConfig=n=1, r=16, %Param=0.122026.058.4
OFTConfig=b=64, %Param=0.122026.058.39
LLaMA2-7BTraining Dataset=Base Model, Model=LLaMA2-7B2024.058.22
TAIATraining Dataset=CoT-Collection, Model=LLaMA2-7B, FT Method=MOLORA, Infer Mode=TAIA2024.058.04
LoRATraining Dataset=Alpaca-GPT4, Model=Qwen1.5-1.8B, FT Method=LoRA, Infer Mode=Vanilla2024.058.02
TAIATraining Dataset=Alpaca-GPT4, Model=LLaMA2-7B, FT Method=LoRA, Infer Mode=TAIA2024.058.02
LoRATraining Dataset=CoT-Collection, Model=LLaMA2-7B, FT Method=LoRA, Infer Mode=Vanilla2024.057.98
MOLORATraining Dataset=CoT-Collection, Model=Qwen1.5-1.8B, FT Method=MOLORA, Infer Mode=Vanilla2024.057.9
LoRATraining Dataset=CoT-Collection, Model=Qwen1.5-1.8B, FT Method=LoRA, Infer Mode=Vanilla2024.057.68
PiSSAPEFT Method=PiSSA, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.087.6
Full FTPEFT Method=Full Fine-tuning, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.087.48
HRAConfig=r=32, %Param=0.122026.057.45
MOLORATraining Dataset=Alpaca-GPT4, Model=LLaMA2-7B, FT Method=MOLORA, Infer Mode=Vanilla2024.057.42
LoRA+PEFT Method=LoRA+, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.087.21
LoRA-GAPEFT Method=LoRA-GA, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.087.13
LoRATraining Dataset=Alpaca-GPT4, Model=LLaMA2-7B, FT Method=LoRA, Infer Mode=Vanilla2024.057.08
BOFTConfig=m=2, b=8, %Param=0.132026.056.85
MiLoRAPEFT Method=MiLoRA, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.086.31
AdaLoRAPEFT Method=AdaLoRA, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.086.24
DoRAPEFT Method=DoRA, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.086.2
LoRAPEFT Method=LoRA, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.085.92
LoRA-FAPEFT Method=LoRA-FA, Base Model=LLaMA-2-7B, Fine-tuning Dataset=MetaMathQA2024.085.66
MOLORATraining Dataset=CoT-Collection, Model=LLaMA2-7B, FT Method=MOLORA, Infer Mode=Vanilla2024.054.54
Qwen1.5-1.8BTraining Dataset=Base Model, Model=Qwen1.5-1.8B2024.054.28
Llama-2-7B2026.052.5