Translation pending. Opens in a new tab
Translation pending.
Thread groupedOpen source Opens in a new tab
River
Primary-source items first added here within the past 48 hours.
Translation pending.
Translation pending.
Translation pending.
Translation pending.
Comment on [版本跟踪] DSH dsh-v0.1.5-rc.2 → dsh-v0.1.7-rc.1
Translation pending.
View thread progress · 2
- Comment on [版本跟踪] DSH dsh-v0.1.5-rc.2 → dsh-v0.1.7-rc.1
Translation pending.
- Translation pending.
Translation pending.
- Comment on [版本跟踪] DSH dsh-v0.1.5-rc.2 → dsh-v0.1.7-rc.1
Translation pending.
Translation pending.
Benchmark Run Overview: Model × Harness × Condition Matrix (S+H Full Series)
This issue indexes benchmark batch results for the model × harness × condition matrix as of upstream/main 02e1404 (2026-09-13), split into official snapshots, DRAFT/pilot, and in-progress tiers. It is only an index and matrix supplement; all figures must come from the original reports.
6 new replies · lhh010
View thread progress · 6
- Comment on Benchmark 跑批结果总览：模型 × harness × 条件矩阵（S+H 全系列）
Translation pending.
- Comment on Benchmark 跑批结果总览：模型 × harness × 条件矩阵（S+H 全系列）
Translation pending.
- Comment on Benchmark 跑批结果总览：模型 × harness × 条件矩阵（S+H 全系列）
Translation pending.
- Comment on Benchmark 跑批结果总览：模型 × harness × 条件矩阵（S+H 全系列）
Translation pending.
- Comment on Benchmark 跑批结果总览：模型 × harness × 条件矩阵（S+H 全系列）
Translation pending.
- Benchmark Run Overview: Model × Harness × Condition Matrix (S+H Full Series)
This issue indexes benchmark batch results for the model × harness × condition matrix as of upstream/main 02e1404 (2026-09-13), split into official snapshots, DRAFT/pilot, and in-progress tiers. It is only an index and matrix supplement; all figures must come from the original reports.
- Comment on Benchmark 跑批结果总览：模型 × harness × 条件矩阵（S+H 全系列）
Proposal: Use a Weaker Model to Restore Benchmark Discrimination
A contributor proposes paired evaluation with deepseek-v4-flash to restore discrimination on a near-saturated benchmark, adding time and cost dimensions. The issue is open with no maintainer response yet.
View thread progress · 2
- Comment on 提案：用较弱模型（deepseek-v4-flash）做配对评测，恢复基准区分度并补充时间/成本维度
Translation pending.
- Proposal: Use a Weaker Model to Restore Benchmark Discrimination
A contributor proposes paired evaluation with deepseek-v4-flash to restore discrimination on a near-saturated benchmark, adding time and cost dimensions. The issue is open with no maintainer response yet.
- Comment on 提案：用较弱模型（deepseek-v4-flash）做配对评测，恢复基准区分度并补充时间/成本维度
DSH v0.1.5-rc.1 → rc.2 version corridor claimed
A contributor claimed the DSH v0.1.5-rc.1 → rc.2 version corridor and is preparing a migration card. The issue pins rc.1's exact commit and npm latest status.
View thread progress · 2
- Comment on [版本跟踪] DSH dsh-v0.1.5-rc.1 → dsh-v0.1.5-rc.2 走廊认领
Translation pending.
- DSH v0.1.5-rc.1 → rc.2 version corridor claimed
A contributor claimed the DSH v0.1.5-rc.1 → rc.2 version corridor and is preparing a migration card. The issue pins rc.1's exact commit and npm latest status.
- Comment on [版本跟踪] DSH dsh-v0.1.5-rc.1 → dsh-v0.1.5-rc.2 走廊认领