来源:paper:pp. 8–9, §4.4, Table 4
这条 Claim 暂无已发布的不可变 Assessment。
这条 Claim 暂无关联执行。
已评估 0/16 项
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。
| 指标与条件 | 论文报告 | 最近可用实测 | 证据判断 |
|---|---|---|---|
variant: w/o Training · category: overall · benchmark: locomo | 48.59 F1 points (0–100) | — | 尚未评估 |
variant: w/o Training · category: overall · benchmark: longmemeval | 48.8% | — | 尚未评估 |
variant: w/o Training · category: overall · benchmark: narrativeqa | 34.98 F1 points (0–100) | — | 尚未评估 |
variant: w/o Training · category: overall · benchmark: hotpotqa | 45.44 F1 points (0–100) | — | 尚未评估 |
variant: SFT Only · category: overall · benchmark: locomo | 50.7 F1 points (0–100) | — | 尚未评估 |
variant: SFT Only · category: overall · benchmark: longmemeval | 60.2% | — | 尚未评估 |
variant: SFT Only · category: overall · benchmark: narrativeqa | 41.59 F1 points (0–100) | — | 尚未评估 |
variant: SFT Only · category: overall · benchmark: hotpotqa | 52.08 F1 points (0–100) | — | 尚未评估 |
variant: GRPO w/o Hint · category: overall · benchmark: locomo | 51.42 F1 points (0–100) | — | 尚未评估 |
variant: GRPO w/o Hint · category: overall · benchmark: longmemeval | 62.4% | — | 尚未评估 |
variant: GRPO w/o Hint · category: overall · benchmark: narrativeqa | 43.62 F1 points (0–100) | — | 尚未评估 |
variant: GRPO w/o Hint · category: overall · benchmark: hotpotqa | 55.64 F1 points (0–100) | — | 尚未评估 |