List your company
ResearchFormerly WizWand

Downstream Repair on Arithmetic trace-level 60/20/20 (test)

92.4Repair AccuracyCROP
62.7670.45578.1585.845May 28, 2026
Updated 5mo ago

Evaluation results

MethodLinks
CROPRepair model=Qwen2.5-7B-Instruct, Alpha=0.05, Input Strategy=CROP prefix2026.0592.41.020.28
Full traceRepair model=Qwen2.5-7B-Instruct, Alpha=0.05, Input Strategy=Full trace2026.0590.8——
Whole-trace abst.Repair model=Llama3.1-8B, Alpha=0.05, Input Strategy=Whole-trace abst.2026.0590.5——
Whole-trace abst.Repair model=Qwen2.5-7B-Instruct, Alpha=0.05, Input Strategy=Whole-trace abst.2026.0590.3——
CROPRepair model=Llama3.1-8B, Alpha=0.05, Input Strategy=CROP prefix2026.0589.4-1.18-1.84
CROPRepair model=Gemma 4 E4B IT, Alpha=0.05, Input Strategy=CROP prefix2026.0588.75.324.41
Full traceRepair model=Llama3.1-8B, Alpha=0.05, Input Strategy=Full trace2026.0587.3——
Question onlyRepair model=Qwen2.5-7B-Instruct, Alpha=0.05, Input Strategy=Question only2026.0585.5——
CROPRepair model=DeepSeek-R1-0528-Qwen3-8B, Alpha=0.05, Input Strategy=CROP prefix2026.05855.534.79
Whole-trace abst.Repair model=Gemma 4 E4B IT, Alpha=0.05, Input Strategy=Whole-trace abst.2026.0583.4——
Question onlyRepair model=Gemma 4 E4B IT, Alpha=0.05, Input Strategy=Question only2026.0580.5——
Full traceRepair model=Gemma 4 E4B IT, Alpha=0.05, Input Strategy=Full trace2026.0579.9——
Whole-trace abst.Repair model=DeepSeek-R1-0528-Qwen3-8B, Alpha=0.05, Input Strategy=Whole-trace abst.2026.0578.8——
Full traceRepair model=DeepSeek-R1-0528-Qwen3-8B, Alpha=0.05, Input Strategy=Full trace2026.0578.5——
Question onlyRepair model=Llama3.1-8B, Alpha=0.05, Input Strategy=Question only2026.0573.7——
Question onlyRepair model=DeepSeek-R1-0528-Qwen3-8B, Alpha=0.05, Input Strategy=Question only2026.0563.9——