SGLang DeepSeek V3/R1 Optimization

Overview

This skill covers the DeepSeek V3/R1 optimization ladder that is active in SGLang main. It intentionally excludes the V3.1 parser delta and the V3.2 DSA/NSA sparse-attention stack, which have separate skills.

Current-main snapshot:

• SGLang origin/main: 929e00eea on 2026-04-21
• sgl-cookbook origin/main: 8ec4d03 on 2026-04-21
• active runtime entry: python/sglang/srt/models/deepseek_v2.py
• DeepSeek V3/R1 entry class: DeepseekV3ForCausalLM
• NextN/MTP entry class: DeepseekV3ForCausalLMNextN

The historical evidence lives in:

• references/pr-history.md: chronological PR evidence and code-level notes
• references/playbook.md: investigation order, symptom mapping, validation commands