Update README.md

README.md

@@ -60,7 +60,7 @@ Whether the workload involves **long-context understanding** or **extended text
 </div>
 
 ### Quickstart
-#### SGLang
+#### SGLang (Recommended)
 ##### Environment Preparation
 ```bash
 pip install uv
@@ -70,7 +70,7 @@ uv venv ~/my_ling_env
 source ~/my_ling_env/bin/activate
 
 # uv pip "sglang-kernel>=0.4.1"
-uv pip install "sglang>=0.5.10.post1"
+uv pip install "sglang[all]>=0.5.10.post1" --prerelease=allow
 ```
 
 ##### Run Inference
@@ -86,6 +86,8 @@ python -m sglang.launch_server \
     --pp-size 1 \
     --dp-size 1 \
     --trust-remote-code \
+    --tool-call-parser qwen25 \
+    --json-model-override-args '{"linear_backend": "seg_la", "torch_dtype": "bfloat16", "architectures": ["BailingMoeV2_5ForCausalLM"], "model_type": "bailing_hybrid", "rope_scaling": {"rope_type": "yarn", "factor": 2.0, "rope_theta": 6000000, "partial_rotary_factor": 0.5, "original_max_position_embeddings": 131072}}' \
     --dist-init-addr $MASTER_IP:2345 \
     --port $PORT \
     --nnodes 1
@@ -116,6 +118,8 @@ python -m sglang.launch_server \
     --mem-fraction-static 0.75 \
     --max-running-requests 64 \
     --max-mamba-cache-size 256 \
+    --tool-call-parser qwen25 \
+    --json-model-override-args '{"linear_backend": "seg_la", "torch_dtype": "bfloat16", "architectures": ["BailingMoeV2_5ForCausalLM"], "model_type": "bailing_hybrid", "rope_scaling": {"rope_type": "yarn", "factor": 2.0, "rope_theta": 6000000, "partial_rotary_factor": 0.5, "original_max_position_embeddings": 131072}}' \
     --trust-remote-code \
     --dist-init-addr $MASTER_IP:2345 \
     --port $PORT \