minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 13 Packages Projects Releases Wiki Activity

[codex] docs: note H200 DeepSeek-V4 checkpoint (#23628)

Browse Source
This commit is contained in:
zijiexia
2026-04-24 00:06:30 -07:00
committed by GitHub
parent 4cb0c4e1f3
commit 1a37e57fb1
1 changed files with 4 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs_new/cookbook/autoregressive/DeepSeek/DeepSeek-V4.mdx
+4
View File
@@ -99,6 +99,10 @@ Please refer to the [official SGLang installation guide](../../../docs/get-start
SGLang supports three main serving recipes for DeepSeek-V4 with different latency/throughput trade-offs (`low-latency`, `balanced`, `max-throughput`), plus specialized recipes for long-context (`cp`, prefill context-parallel) and prefill/decode disaggregation (`pd-disagg`). The interactive generator below emits the exact launch command for any `(hardware, variant, recipe)` combination. SGLang supports three main serving recipes for DeepSeek-V4 with different latency/throughput trade-offs (`low-latency`, `balanced`, `max-throughput`), plus specialized recipes for long-context (`cp`, prefill context-parallel) and prefill/decode disaggregation (`pd-disagg`). The interactive generator below emits the exact launch command for any `(hardware, variant, recipe)` combination.
<Note>
For H200 GPU deployments, use the SGLang checkpoint under `sgl-project`, not the default DeepSeek checkpoint.
</Note>
### 3.1 Basic Configuration ### 3.1 Basic Configuration
**Interactive Command Generator**: Use the selector below to generate the deployment command for your hardware + recipe combination. **Interactive Command Generator**: Use the selector below to generate the deployment command for your hardware + recipe combination.