For untrained JITMEM-base, storing all trajectories with correctness labels rather than filtering to successful trajectories lowers success by 1.5-2.9 points on ALFWorld and 2.3-3.4 on WebShop across executors; replacing raw traces with write-time ReasoningBank-style distillations lowers success by 1.7-2.9 on ALFWorld and 6.8-8.2 on WebShop.

For untrained JITMEM-base, storing all trajectories with correctness labels rather than filtering to successful trajectories lowers success by 1.5-2.9 points on ALFWorld and 2.3-3.4 on WebShop across executors; replacing raw traces with write-time ReasoningBank-style distillations lowers success by 1.7-2.9 on ALFWorld and 6.8-8.2 on WebShop. · CiteArk

测量结果

已评估 0/18 项

获得支持 0有争议或冲突 0尚未评估 18

保留同一结论下的完整结果组。已有评估排在前面；单项判断不替代整组结论。展开行查看条件、来源与历次证据。

每行是一项测量，分别显示论文报告、最近可用实测和累计证据状态。
论文报告	最近可用实测	证据判断
59 percentage points	—	尚未评估
23.5 points	—	尚未评估
8.9 percentage points	—	尚未评估
77.1 percentage points	—	尚未评估
52 points	—	尚未评估
42.1 percentage points	—	尚未评估
77.6 percentage points	—	尚未评估
45.5 points	—	尚未评估
35.9 percentage points	—	尚未评估
58.8 percentage points	—	尚未评估
17.8 points	—	尚未评估
4.9 percentage points	—	尚未评估

报告指标与观测值

Assessment（0）

关联运行（0）

测量结果