At the same final checkpoints used for the matched-compute comparison, 67B SST scores higher than both classic baselines on all seven reported downstream tasks and has the lowest held-out ARXIV token NLL: 1.4421 versus 1.4424 for 63B classic and 1.4521 for 47B classic.

At the same final checkpoints used for the matched-compute comparison, 67B SST scores higher than both classic baselines on all seven reported downstream tasks and has the lowest held-out ARXIV token NLL: 1.4421 versus 1.4424 for 63B classic and 1.4521 for 47B classic. · CiteArk

测量结果

已评估 0/24 项

获得支持 0有争议或冲突 0尚未评估 24

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。
指标与条件	论文报告	最近可用实测	证据判断
model: 47B classic · split: validation + test	68.5%	—	尚未评估
model: 63B classic · split: validation + test	71%	—	尚未评估
model: 67B SST · split: validation + test	75%	—	尚未评估
model: 47B classic · split: test	58.44%	—	尚未评估
model: 63B classic · split: test	59.39%	—	尚未评估
model: 67B SST · split: test	60.54%	—	尚未评估
model: 47B classic	56.56%	—	尚未评估
model: 63B classic	55.04%	—	尚未评估
model: 67B SST	58.38%	—	尚未评估
model: 47B classic · split: full test	33.06%	—	尚未评估
model: 63B classic · split: full test	33.14%	—	尚未评估
model: 67B SST · split: full test	34.36%	—	尚未评估

At the same final checkpoints used for the matched-compute comparison, 67B SST scores higher than both classic baselines on all seven reported downstream tasks and has the lowest held-out ARXIV token NLL: 1.4421 versus 1.4424 for 63B classic and 1.4521 for 47B classic.

报告指标与观测值

Assessment（0）

关联运行（0）

测量结果