List your company

Multi-step Reasoning on GSM8K (test)

32.2Pass@1SLR
-0.1448.25316.6525.047Feb 6, 2026

Evaluation results

MethodLinks
SLRBackbone=Qwen2026.0232.26071.679.48590.5
Proxy-SoupBackbone=Qwen2026.022852.562.368.973.676.8
SLRBackbone=Gemma2026.0221.245.65970.779.184.2
SLRBackbone=Llama2026.0215.442.961.277.888.994.7
Proxy-SoupBackbone=Llama2026.0213.637.552.766.375.980
Proxy-SoupBackbone=Gemma2026.028.320.529.537.642.344.2
Post-trainedBackbone=Qwen2026.027.42540.456.366.772.6
Post-trainedBackbone=Llama2026.024.516.228.2445866.3
Post-trainedBackbone=Gemma2026.021.147.51319.926.3