List your company

Multi-table Reasoning on BIRD 3-table

56.25AccuracyCPO
17.2527.37537.547.625Jun 3, 2026

Evaluation results

MethodLinks
CPOModel=Qwen3-14B2026.0656.25
CPOModel=Llama-3.1-8B2026.0656.25
Q&A SFTModel=Qwen3-14B2026.0650
Trace SFTModel=Qwen3-14B2026.0650
CPOModel=Mistral-8B2026.0643.75
Trace SFTModel=Llama-3.1-8B2026.0643.75
Q&A SFTModel=Mistral-8B2026.0631.25
Q&A SFTModel=Llama-3.1-8B2026.0631.25
Trace SFTModel=Mistral-8B2026.0618.75