Knowledge Distillation During Mid-Training Favors Reasoning over Factual Recall
尚未评估计划受阻指标结果claim_t3_random_routing

Mid-training ablation results for Random Routing relative difference across Reasoning, Factual Recall, and Knowledge task groups.

来源:source_paper:Table 3, Page 9

报告指标与观测值

delta_percentage_points

t3_random_routing_reasoning

论文报告 -6.5 percentage_points

实际观测 — percentage_points

delta_percentage_points

t3_random_routing_factual_recall

论文报告 -0.8 percentage_points

实际观测 — percentage_points

delta_percentage_points

t3_random_routing_knowledge

论文报告 -2 percentage_points

实际观测 — percentage_points

Assessment(0)

这条 Claim 暂无已发布的不可变 Assessment。

关联运行(0)

这条 Claim 暂无关联执行。