wuyang6
18 lines
554 B
Markdown
18 lines
554 B
Markdown
---
|
|
{
|
|
"title": "27B 推理优化：从基线到生产配置",
|
|
"summary": "记录 27B 模型在 llama.cpp 路径上的方案变化、三项速度指标和生产配置取舍。",
|
|
"date": "2026-07-28",
|
|
"updated": "2026-08-10",
|
|
"topic": "llm-systems",
|
|
"tags": ["inference", "llama.cpp", "optimization"],
|
|
"kind": "site",
|
|
"status": "published",
|
|
"visibility": "public",
|
|
"canonicalUrl": "https://llama-cpp-report.k1412.top",
|
|
"preview": "/previews/llama-cpp-report.jpg"
|
|
}
|
|
---
|
|
|
|
独立报告站。历史源码尚未迁入中央公开仓库。
|