List your company

Program-aided math reasoning on TabMWP

75.3AccuracyDeepSeek-Coder-Base
28.18840.41952.6564.881Jan 25, 2024

Evaluation results

MethodLinks
DeepSeek-Coder-BaseSize=33B, Protocol=Program-Aided Math Reasoning (PAL)2024.0175.3
CodeLlama-BaseSize=34B, Protocol=Program-Aided Math Reasoning (PAL)2024.0169.8
DeepSeek-Coder-BaseSize=6.7B, Protocol=Program-Aided Math Reasoning (PAL)2024.0167.9
CodeLlama-BaseSize=13B, Protocol=Program-Aided Math Reasoning (PAL)2024.0160.3
CodeLlama-BaseSize=7B, Protocol=Program-Aided Math Reasoning (PAL)2024.0152.9
StarCoder-BaseSize=16B, Protocol=Program-Aided Math Reasoning (PAL)2024.0145
CodeGeex-2Size=7B, Protocol=Program-Aided Math Reasoning (PAL)2024.0144.6
DeepSeek-Coder-BaseSize=1.3B, Protocol=Program-Aided Math Reasoning (PAL)2024.0130