List your company

Mathematical Reasoning on GSM-8K (GSM, URR)

84.8GSM AccuracyQwen2.5-7B-Instruct
15.74433.67251.669.528Apr 14, 2026

Evaluation results

MethodLinks
Qwen2.5-7B-InstructBase Model=Qwen2.5-7B-Instruct2026.0484.8100
DualEditBase Model=Qwen2.5-7B-Instruct2026.0482.886.2
STEEREDITBase Model=Qwen2.5-7B-Instruct2026.048297.2
ROMEBase Model=Qwen2.5-7B-Instruct2026.0480.197.7
SFTBase Model=Qwen2.5-7B-Instruct2026.0479.392.2
Llama-3-8B-InstructBase Model=Llama-3-8B-Instruct2026.0474.6100
STEEREDITBase Model=Llama-3-8B-Instruct2026.047498.6
DualEditBase Model=Llama-3-8B-Instruct2026.0473.895.7
JailbreakEditBase Model=Qwen2.5-7B-Instruct2026.0472.769.3
ROMEBase Model=Llama-3-8B-Instruct2026.0465.685.9
JailbreakEditBase Model=Llama-3-8B-Instruct2026.0464.670.1
SFTBase Model=Llama-3-8B-Instruct2026.0448.881.2
Llama-2-7b-chatBase Model=Llama-2-7b-chat2026.0422.7100
SFTBase Model=Llama-2-7b-chat2026.0422.386.7
ROMEBase Model=Llama-2-7b-chat2026.0421.190.9
STEEREDITBase Model=Llama-2-7b-chat2026.042197
DualEditBase Model=Llama-2-7b-chat2026.0420.995.7
JailbreakEditBase Model=Llama-2-7b-chat2026.0418.470.7