Update README.md (#1)

- Update README.md (71af1b053821e846ee311462d9b28308c7f823a1)


Co-authored-by: Yue yu <vitoyy@users.noreply.huggingface.co>

README.md

@@ -58,23 +58,25 @@ docker pull lmsysorg/sglang:dev-Ling-3.0-flash-VL
 ```
 
 ### Run Inference
-Recommended recipe with 256K context (YaRN), on 4× 141GB-class GPUs (H20-3e / H200) or 4-GPU Blackwell nodes (B300 / GB300):
+Recommended recipe with 256K context (YaRN), on 2× 141GB-class GPUs (H20-3e / H200) or 2-GPU Blackwell nodes (B300 / GB300):
 
 ```bash
 docker run --rm --gpus all --ipc=host --shm-size 32g \
   -p 30000:30000 \
   -e HF_TOKEN=<your-hf-token> \
+  -e SGLANG_ALLOW_OVERWRITE_LONGER_CONTEXT_LEN=1 \
   lmsysorg/sglang:dev-Ling-3.0-flash-VL \
-  env SGLANG_ALLOW_OVERWRITE_LONGER_CONTEXT_LEN=1 \
   python3 -m sglang.launch_server \
-    --model-path inclusionAI/Ling-3.0-flash-VL \
-    --tp 4 \
+    --model-path inclusionAI/Ling-3.0-flash-VL-int4 \
+    --tp-size 2 \
     --context-length 262144 \
     --json-model-override-args '{"rope_scaling":{"rope_type":"yarn","factor":2.0,"rope_theta":6000000,"partial_rotary_factor":0.5,"original_max_position_embeddings":131072}}' \
     --mem-fraction-static 0.85 \
     --trust-remote-code \
     --reasoning-parser auto \
     --tool-call-parser auto \
+    --enable-fp32-lm-head \
+    --disable-shared-experts-fusion \
     --host 0.0.0.0 \
     --port 30000
 ```