List your company

Mathematical Reasoning on GSM8K (Acc, Len, CR)

96.2AccuracyASAG
85.90488.57791.2593.923Jun 13, 2026

Evaluation results

MethodLinks
ASAGBackbone=Qwen3-14B2026.0696.21,03949.9
ASAGBackbone=Qwen3-8B2026.0695.91,06345.5
DynasorBackbone=Qwen3-14B2026.0695.61,36465.5
DEERBackbone=Qwen3-14B2026.0695.386341.5
VanillaBackbone=Qwen3-14B2026.0695.12,082100
DEERBackbone=Qwen3-8B2026.06951,09646.9
TALEBackbone=Qwen3-14B2026.0694.796346.3
VanillaBackbone=Qwen3-8B2026.0694.62,338100
DEERBackbone=Qwen3-4B2026.0694.21,28660
ASAGBackbone=Qwen3-4B2026.0694.21,13953.2
DynasorBackbone=Qwen3-8B2026.0694.21,02443.8
VanillaBackbone=Qwen3-4B2026.0693.82,142100
NoThinkingBackbone=Qwen3-14B2026.0693.544721.5
DynasorBackbone=Qwen3-4B2026.0692.993843.8
TALEBackbone=Qwen3-8B2026.0692.51,14248.8
TALEBackbone=Qwen3-4B2026.0691.392643.2
NoThinkingBackbone=Qwen3-8B2026.0691.368529.3
ASAGBackbone=DeepSeek-R1-Distill-Qwen-7B2026.06911,00969.5
DEERBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0690.396766.6
DynasorBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0689.71,25886.6
NoThinkingBackbone=Qwen3-4B2026.0689.662429.1
VanillaBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0689.51,452100
TALEBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0688.290962.6
NoThinkingBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0686.332822.6