The gap between ASPIRE and synchronized ASPIRE-Fixed was small on short-context tasks but grew with context length. On Qwen3-8B, their speedups were 1.81× versus 1.73× at 16k–18k, 2.30× versus 1.98× at 30k–40k, and 2.75× versus 2.22× at 80k–100k; on DS-Llama-8B they were 2.49× versus 2.12× and 2.66× versus 2.21× on the two LongBench-v2 ranges.

The gap between ASPIRE and synchronized ASPIRE-Fixed was small on short-context tasks but grew with context length. On Qwen3-8B, their speedups were 1.81× versus 1.73× at 16k–18k, 2.30× versus 1.98× at 30k–40k, and 2.75× versus 2.22× at 80k–100k; on DS-Llama-8B they were 2.49× versus 2.12× and 2.66× versus 2.21× on the two LongBench-v2 ranges. · CiteArk

测量结果

已评估 0/10 项

获得支持 0有争议或冲突 0尚未评估 10

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: Qwen3-8B · method: ASPIRE · dataset: LB[16k-18k]	1.81 × AutoRegressive	—	尚未评估
model: Qwen3-8B · method: ASPIRE-Fixed · dataset: LB[16k-18k]	1.73 × AutoRegressive	—	尚未评估
model: Qwen3-8B · method: ASPIRE · dataset: LB-v2[30k-40k]	2.3 × AutoRegressive	—	尚未评估
model: Qwen3-8B · method: ASPIRE-Fixed · dataset: LB-v2[30k-40k]	1.98 × AutoRegressive	—	尚未评估
model: Qwen3-8B · method: ASPIRE · dataset: LB-v2[80k-100k]	2.75 × AutoRegressive	—	尚未评估
model: Qwen3-8B · method: ASPIRE-Fixed · dataset: LB-v2[80k-100k]	2.22 × AutoRegressive	—	尚未评估
model: DS-Llama-8B · method: ASPIRE · dataset: LB-v2[30k-40k]	2.49 × AutoRegressive	—	尚未评估
model: DS-Llama-8B · method: ASPIRE-Fixed · dataset: LB-v2[30k-40k]	2.12 × AutoRegressive	—	尚未评估
model: DS-Llama-8B · method: ASPIRE · dataset: LB-v2[80k-100k]	2.66 × AutoRegressive	—	尚未评估
model: DS-Llama-8B · method: ASPIRE-Fixed · dataset: LB-v2[80k-100k]	2.21 × AutoRegressive	—	尚未评估

报告指标与观测值

Assessment（0）

关联运行（0）

测量结果