一个数据流向图提示词：用箭头粗细编码数据量和类型
来源：GitHub 仓库 XYD-GIS/paper-toolkits 文件约 85 行，约 3.8 KB
一个数据流向图提示词。
它解决的是：要画"数据如何在系统中流转"的图，模块本身次要，箭头的粗细/颜色编码数据类型与体量。
它的结构：
- 布局：从左到右横向流动
- 节点：4-6 个圆角矩形，每个有英文标签（1-3 词）和小图标提示
- 箭头：细箭头（1-2px）表元数据/控制信号，中箭头（3-4px）表特征向量，粗箭头（6-8px）表原始数据/高维张量，统一灰色
- 标注：每条箭头上方用小号斜体灰字标数据类型
- 可选分支：用虚线箭头
- 风格：扁平矢量、白底、IEEE/ACM 图审美、Arial、16:9
- 负向约束：不要照片、不要 3D、不要卡通、不要重阴影、不要难读文字、不要混乱交叉箭头
填空示例（视觉分类系统）：
- 节点列表 = Raw Data → Preprocessor → Encoder → Classifier → Output
- 图标题 = Data Flow in Vision Classifier
- 具体标注：
- Raw Data → Preprocessor: thick arrow, label "RGB image (224×224×3)"
- Preprocessor → Encoder: medium arrow, label "normalized tensor"
- Encoder → Classifier: medium arrow, label "embedding (768-d)"
- Classifier → Output: thin arrow, label "logits → softmax → label"
调优提示：
- 有反馈支路 → 加 "Plus a dashed feedback arrow from Output back to Encoder, labeled 'gradient' in light gray"
- 数据量类型超过 3 种 → 可以用颜色而不是粗细编码，但保持色盲友好
- 想强调瓶颈点 → 在某条箭头旁加 "label this arrow with a small red badge '⚠ bottleneck'"
适合谁：
- 写论文要画数据流向图的科研小组成员
- 需要强调数据流转的人
- 想要 IEEE/ACM 图审美的人
用法： 把这段提示词复制到 AI 工具里，把占位符填上就行。
提示词
A data flow diagram for an academic paper, focused on tracing how information moves through a system. The diagram uses arrows of varying thickness to represent data volume / importance.
LAYOUT: Horizontal flow from left to right.
NODES (rounded rectangles, 4-6 total):
{节点列表，如 "Raw Data → Preprocessor → Encoder → Classifier → Output"}
Each node has an English label (1-3 words) and a small icon hint inside.
ARROWS:
- Thin arrows (1-2 px): represent metadata or control signal
- Medium arrows (3-4 px): represent feature vectors
- Thick arrows (6-8 px): represent raw data or high-volume tensors
- Use the same gray color (#6B7280) for all arrows; vary only thickness.
ANNOTATIONS:
- Above each arrow, in small italic gray text, label the data type (e.g., "RGB image", "embedding 768-d", "class probabilities").
- At the top center, place a clean title in bold Arial: "{图标题}"
OPTIONAL BRANCHES: If the flow has any side branches (e.g., auxiliary loss, residual connection), draw them as dashed arrows in a lighter gray.
Style: flat vector, white background, IEEE / ACM figure aesthetic, Arial sans-serif. Aspect ratio 16:9.
Negative constraints: NO photorealistic, NO 3D, NO cartoon, NO heavy shadows, NO unreadable text, NO chaotic crossing arrows.
# 填空示例
{节点列表} = Raw Data → Preprocessor → Encoder → Classifier → Output
{图标题} = Data Flow in Vision Classifier
SPECIFIC ANNOTATIONS:
- Raw Data → Preprocessor: thick arrow, label "RGB image (224×224×3)"
- Preprocessor → Encoder: medium arrow, label "normalized tensor"
- Encoder → Classifier: medium arrow, label "embedding (768-d)"
- Classifier → Output: thin arrow, label "logits → softmax → label"评论 0
更多
登录后可点赞、收藏、评论和举报。
还没有评论，先发起一个具体问题。