List your company
ResearchFormerly WizWand

Instruction Following on IFEval (Score %)

85.4Score (%)Qwen3-8B
25.8641.317556.77572.2325May 16, 2026May 22, 2026May 29, 2026Jun 5, 2026Jun 11, 2026Jun 18, 2026Jun 25, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Qwen3-8Binference_mode=thinking mode2026.0685.4
Teacher Bestinference_mode=thinking mode2026.0684.8
MOPDinference_mode=thinking mode2026.0677.26
Base (SFT)inference_mode=thinking mode2026.0663.77
Task ArithmeticModel Scale=Llama-3.1-3B, Quantization Bit-width=Full-precision, Merging Algorithm=Task Arithmetic2026.0562.48
E-PMQModel Scale=Llama-3.1-3B, Quantization Bit-width=4-bit, Merging Algorithm=Task Arithmetic2026.0562.48
AWQModel Scale=Llama-3.1-3B, Quantization Bit-width=4-bit, Merging Algorithm=Task Arithmetic2026.0560.63
GPTQModel Scale=Llama-3.1-3B, Quantization Bit-width=4-bit, Merging Algorithm=Task Arithmetic2026.0558.78
ARES-RL2026.0554.88
ARES-RLBackbone=Qwen3-32B2026.0554.88
Task ArithmeticModel Scale=Llama-3.1-8B, Quantization Bit-width=Full-precision, Merging Algorithm=Task Arithmetic2026.0550.09
E-PMQModel Scale=Llama-3.1-8B, Quantization Bit-width=4-bit, Merging Algorithm=Task Arithmetic2026.0548.8
ARES-SFT2026.0546.41
ARES-SFTBackbone=Qwen3-32B2026.0546.41
GPTQModel Scale=Llama-3.1-8B, Quantization Bit-width=4-bit, Merging Algorithm=Task Arithmetic2026.0544.92
AWQModel Scale=Llama-3.1-8B, Quantization Bit-width=4-bit, Merging Algorithm=Task Arithmetic2026.0544.73
CPT2026.0539.39
CPTBackbone=Qwen3-32B2026.0539.39
Webscale2026.0535.61
WebscaleBackbone=Qwen3-32B2026.0535.61
Natural Reasoning2026.0528.15
NaturalReasoningBackbone=Qwen3-32B2026.0528.15