|
|
@@ -15,13 +15,11 @@ const THREE_POINT_TWO_TWO_CHANGELOG: ChangelogEntry = {
|
|
|
"[Qwen3.5/3.6 Ollama 500 Fix] Extra system messages for Qwen3.5/3.6 are merged into the first one, so local Ollama chapter writing and outline generation no longer fail with HTTP 500 (System message must be at the beginning).",
|
|
|
"[Re-extract Overwrites Timeline] Re-extracting chapter memory now replaces that chapter's tracking timeline instead of appending, so the same events no longer pile up as duplicate rows.",
|
|
|
"[Faster Chapter Memory Extract] Chapter ingest uses fewer tokens and skips duplicate wiki/snapshot writes; re-extract rebuilds derived memory from snapshots instead of stacking old foreshadowing and cognition.",
|
|
|
- "[Fast Writing Uses Selected Model] Fast-mode chat/output now uses the writing model selected in the chat picker, not the default main-agent orchestration model.",
|
|
|
],
|
|
|
zh: [
|
|
|
"【Qwen3.5/3.6 Ollama 500 修复】本地 Ollama 用 Qwen3.5/3.6 写章节或大纲时,多余的 system 会合到第一条,不再报 HTTP 500(System message must be at the beginning)",
|
|
|
"【重提取覆盖时间线】重新提取章节记忆时,该章 tracking 时间线改为覆盖而不是追加,同一批事件不会再堆出重复行",
|
|
|
"【章节记忆提取更快】摄取少耗 token、去掉重复写入;重提取从快照重建派生记忆,不再把旧认知/伏笔叠进去",
|
|
|
- "【快速写作用所选模型】快速模式下对话/出稿改用聊天框选中的写作模型,不再误用默认主 Agent 编排模型",
|
|
|
],
|
|
|
},
|
|
|
};
|