怎么填变量：三段内容可以完全换掉，比如做"公司预算流向"：左段写"[营收来源]"，中段写"研发 / 市场 / 运营"，右段写"利润 / 再投资"；做"用户转化"：左段写各渠道访客，中段写"注册 / 激活"，右段写"付费 / 流失"。数字写成"5400 亿"这类短格式更容易被正确画出来。示例图是英文版：左边八个彩色色块，中间三个白色处理框，右边一大块浅青"Pretraining set"和两块橙色小块，流带在中间交织，底部有图例。

常见问题与调整：

• 流带粗细和数字对不上：把来源控制在 5 个以内，并加"流带粗细严格与数字成比例"。
• 中文标签被流带遮住：加"所有标签放在色块内部，流带上只标数字"。
• 颜色太乱：限定"只用 3 种主色的深浅变化"。
• 需要精确数据：用这张定配色和版式，再用图表工具按真实数据画。

适合：技术分享、年度报告、业务流程讲解 PPT 的示意图；不适合当作精确统计结果发布。

英文原版

Landscape 16:9 sankey diagram of a pretraining data mixture, three stages with translucent colored ribbons.

LEFT (8 source blocks, heights proportional to tokens): "Common Crawl (web) 540B" (muted navy, largest), "arXiv papers 180B" (dusty teal), "GitHub code 160B" (slate gray), "Wikipedia 40B" (soft terracotta), "StackExchange QA 30B" (warm copper), "Books (public domain) 25B" (pale olive), "Patents 18B" (pale navy), "Curated news & forums 15B" (dusty teal).

MIDDLE (3 processing blocks, stacked): "Deduplicated (MinHash + exact)", "Quality-filtered (classifier + heuristics)", "PII-scrubbed (regex + NER)".

RIGHT (3 final splits): "Pretraining set 1.4T tokens" (largest), "Instruction-tune pool 12B tokens", "RLHF preference pool 3B tokens".

Flow ribbons inherit source color with mid-labels showing token counts ("85B", "320B", "44B"). Legend strip at bottom.

Title: "LLM pretraining data mixture and downstream splits". Subtitle: "token counts after deduplication and quality filtering; ribbon thickness ∝ token flow."

改编自 wuyoscar/GPT-Image2-Skill 收录的提示词（仓库整理），许可证 MIT（Copyright (c) 2026 Wuyoscar）。