ResearchBenchmarksMathematical Reasoning on GSM8K (Fine-tuning Accuracy)Follow16.27Fine-tuning Accuracy (FA)Booster8.948410.849212.7514.6508Feb 28, 2026Evaluation ResultsMethodMethodLinksFine-tuning Accuracy (FA)BoosterBackbone=Llama-2-7BBackbone=Llama-2-7B2026.0216.27AntibodyBackbone=Llama-2-7BBackbone=Llama-2-7B2026.0215.07VaccineBackbone=Llama-2-7BBackbone=Llama-2-7B2026.0211.7SFTBackbone=Llama-2-7BBackbone=Llama-2-7B2026.0210.9LisaBackbone=Llama-2-7BBackbone=Llama-2-7B2026.029.23