2 Likes
在语文教科书移除滕王阁序之后，个人感觉考察范围不会再将其包括其中，有没有必要把其从高考默写网站里移除一下。@suen
1 Like
看說明先：）
1 Like
Custom title
By default, the title of the callout is its type identifier in title case. You can change it by adding text after the type identifier:
[!question] Custom Label
> Callouts labels can be customized. > [!abstract] 2032-12-21T23:00:00Z > They can also contain special inline formating like dynamic dates.
You can also create title-only callouts:
[!success] solved.
Foldable callouts
You can make a callout foldable by adding a plus + or a minus - directly after the type identifier.
+expands the callout by default-collapses it instead
[!todo]- My Todo List > > - Write documentation > - Add tests > - Review code > [!todo]+ My Expanded Todo List > > - Create repository > - Setup CI/CD > - Deploy to production
Nested callouts
You can nest callouts in multiple levels.
[!question] Can callouts be nested? > > > [!todo] Yes!, they can. > > > > > [!example] You can even use multiple layers of nesting. > [!tip] > Here’s a helpful tip > > > [!info] > > With some additional information > > > > > [!example] > > > And a nested example > > Back to the main tip
Supported default types
You can use several callout types and aliases. Each type comes with a different background color and icon.
You can create your callouts or manage the default ones. See the
calloutssetting.
notetype. See Default Fallbacks section for more information.
| Type | Description | Aliases |
|---|---|---|
note |
For general notes and information | - |
abstract |
For summaries or abstract sections | summary, tldr |
info |
For informational content | - |
todo |
For task lists or todo items | - |
tip |
For tips and important information | hint, important |
success |
For success messages | check, done |
question |
For questions and FAQs | help, faq |
warning |
For warnings | caution, attention |
failure |
For failure notices | fail, missing |
danger |
For danger or error messages | error |
bug |
For bug reports or known issues | - |
example |
For examples | - |
quote |
For quotes | cite |
3 Likes
例行更新完畢。耗時長了些是因為報錯：
docker: “specify mac-address per network” requires API version 1.44,
but the Docker daemon API version is 1.43.
導致無法啟動。已更新 Docker Engine 到與 CLI 匹配的版本
2 Likes
https://forum.rdfzer.com/discourse-ai/ai-bot/conversations
顧名可以思義，開啟幾個新的人機。
1 Like
刪繁就簡三秋樹：
- 合併之前紅樓/論語為一個版區。
- 合併諸子/史傳為一個版區。
3 Likes
加入一個測試模型：
Pixtral Large
Mistral’s most powerful vision capable model
主要用來結構化OCR。
2 Likes
已完成例行更新。
Gemini識別圖片存在問題，錯誤原因已找到：
no implicit conversion of Array into String
需要等修復，可能要有點時間。
修復之前給Gemini圖片會⋯⋯導致啞巴：）
調整deepseek模型為 DeepSeek: R1 0528 (free)並暫時將官方 API 接口更換為 OpenRouter 。
1 Like
Image recognition can use grok, and we will notify you here after it’s fixed. The grok model has been adjusted to the Gemini model.
1 Like
As long as the post has images, grok it first.
1 Like
Meta’s latest Llama-4-Maverick model has been deployed. Its actual conversational performance is neither good nor bad, and it will be retained.
1 Like
看完官方文檔，數據還是會傳到DS官方，那不如直接官方，已刪除。
1 Like
这个模文学鉴赏上感觉稍逊于2.5flash，但是cheaper（相对于2.5flash reasoning的价格）
2 Likes
AI image recognition features like creative have been restored. The cause has been identified as a code conflict that occurred after enabling RAG. RAG has been temporarily disabled.
Note: The deepseek model inherently cannot recognize images; all others can.
1 Like
已調整 creative 和 suenai 模型為Gemini 2.5 pro。
grok 等其他模型保持為Gemini 2.5 Flash。
體感上，Gemini 2.5 pro模型下的 AI 的回覆速度要慢大約40秒，但質量確實更高。
1 Like
Rich person 
3 Likes
Actually, Gemini 2.5 Pro has a free tier again, and the limit is now 100, which is double the previous 50.
1 Like