List your company

Mathematical Reasoning on GSM8K (T, U, F)

91.4T (Success Rate)Binary
83.39285.47187.5589.629Apr 3, 2026

Evaluation results

MethodLinks
BinaryBackbone=Qwen2.5-7B-Instruct2026.0491.408.6
KARLBackbone=Qwen2.5-7B-Instruct2026.0490.40.49.2
PromptingBackbone=Qwen2.5-7B-Instruct2026.0489.10.510.4
TernaryBackbone=Qwen2.5-7B-Instruct2026.0488.538.5
KARLBackbone=Llama-3.1-8B-Instruct2026.0486.11.312.6
TernaryBackbone=Llama-3.1-8B-Instruct2026.0485.5311.5
BinaryBackbone=Llama-3.1-8B-Instruct2026.0485.2014.8
PromptingBackbone=Llama-3.1-8B-Instruct2026.0483.70.515.8