minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

docs: verify GB300 Pro DeepSeek V4 recipes (#23817)

Browse Source
This commit is contained in:
zijiexia
2026-04-27 00:21:26 -07:00
committed by GitHub
parent 5f47cae1a0
commit c2ec64f243
1 changed files with 6 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs_new/src/snippets/autoregressive/deepseek-v4-deployment.jsx
+6
View File
@@ -182,7 +182,9 @@ export const DeepSeekV4Deployment = () => {
"gb300|small|low-latency", "gb300|small|low-latency",
"gb300|big|low-latency", "gb300|big|low-latency",
"gb300|small|balanced", "gb300|small|balanced",
"gb300|big|balanced",
"gb300|small|max-throughput", "gb300|small|max-throughput",
"gb300|big|max-throughput",
"h200|small|cp", "h200|small|cp",
"h200|small|pd-disagg", "h200|small|pd-disagg",
"h200|big|low-latency", "h200|big|low-latency",
@@ -365,6 +367,8 @@ export const DeepSeekV4Deployment = () => {
flags.push(" --speculative-num-draft-tokens 2"); flags.push(" --speculative-num-draft-tokens 2");
if (hardware === "h200" && isBig) { if (hardware === "h200" && isBig) {
flags.push(" --mem-fraction-static 0.88"); flags.push(" --mem-fraction-static 0.88");
} else if (isBig && hardware === "gb300") {
flags.push(" --mem-fraction-static 0.9");
} else if (isBig && hardware === "gb200") { } else if (isBig && hardware === "gb200") {
flags.push(" --mem-fraction-static 0.78"); flags.push(" --mem-fraction-static 0.78");
} else if (isBig) { } else if (isBig) {
@@ -401,6 +405,8 @@ export const DeepSeekV4Deployment = () => {
flags.push(" --moe-a2a-backend deepep"); flags.push(" --moe-a2a-backend deepep");
if (hardware === "h200" && isBig) { if (hardware === "h200" && isBig) {
flags.push(" --mem-fraction-static 0.88"); flags.push(" --mem-fraction-static 0.88");
} else if (isBig && hardware === "gb300") {
flags.push(" --mem-fraction-static 0.9");
} else if (isBig && hardware === "gb200") { } else if (isBig && hardware === "gb200") {
flags.push(" --mem-fraction-static 0.78"); flags.push(" --mem-fraction-static 0.78");
} else if (isBig) { } else if (isBig) {