update README

docs/sglang_deploy_guide_cn.md

@@ -34,9 +34,12 @@
 
 建议在全新的 Python 环境中安装 SGLang。由于尚未 release，需要从源码手动编译：
 ```bash
-git clone https://github.com/sgl-project/sglang.git
+git clone -b v0.5.4.post3 https://github.com/sgl-project/sglang.git
 cd sglang
-uv pip install ./python --torch-backend=auto
+
+# Install the python packages
+pip install --upgrade pip
+pip install -e "python"
 ```
 
 运行如下命令启动 SGLang 服务器，SGLang 会自动从 Huggingface 下载并缓存 MiniMax-M2 模型。

docs/vllm_deploy_guide.md

@@ -42,6 +42,16 @@ uv pip install vllm --extra-index-url https://wheels.vllm.ai/nightly
 
 Run the following command to start the vLLM server. vLLM will automatically download and cache the MiniMax-M2 model from Hugging Face.
 
+4-GPU deployment command:
+
+```bash
+SAFETENSORS_FAST_GPU=1 vllm serve \
+    MiniMaxAI/MiniMax-M2 --trust-remote-code \
+    --tensor-parallel-size 4 \
+    --enable-auto-tool-choice --tool-call-parser minimax_m2 \
+    --reasoning-parser minimax_m2_append_think
+```
+
 8-GPU deployment command:
 
 ```bash
@@ -49,8 +59,7 @@ SAFETENSORS_FAST_GPU=1 vllm serve \
     MiniMaxAI/MiniMax-M2 --trust-remote-code \
     --enable_expert_parallel --tensor-parallel-size 8 \
     --enable-auto-tool-choice --tool-call-parser minimax_m2 \
-    --reasoning-parser minimax_m2_append_think \
-    --compilation-config "{\"cudagraph_mode\": \"PIECEWISE\"}"
+    --reasoning-parser minimax_m2_append_think 
 ```
 
 ## Testing Deployment
@@ -83,6 +92,18 @@ export HF_ENDPOINT=https://hf-mirror.com
 
 This vLLM version is outdated. Please upgrade to the latest version.
 
+### torch.AcceleratorError: CUDA error: an illegal memory access was encountered
+Add `--compilation-config "{\"cudagraph_mode\": \"PIECEWISE\"}"` to the startup parameters to resolve this issue. For example:
+
+```bash
+SAFETENSORS_FAST_GPU=1 vllm serve \
+    MiniMaxAI/MiniMax-M2 --trust-remote-code \
+    --enable_expert_parallel --tensor-parallel-size 8 \
+    --enable-auto-tool-choice --tool-call-parser minimax_m2 \
+    --reasoning-parser minimax_m2_append_think \
+    --compilation-config "{\"cudagraph_mode\": \"PIECEWISE\"}"
+```
+
 ## Getting Support
 
 If you encounter any issues while deploying the MiniMax model:

docs/vllm_deploy_guide_cn.md

@@ -41,6 +41,16 @@ uv pip install vllm --extra-index-url https://wheels.vllm.ai/nightly
 
 运行如下命令启动 vLLM 服务器，vLLM 会自动从 Huggingface 下载并缓存 MiniMax-M2 模型。
 
+4 卡部署命令：
+
+```bash
+SAFETENSORS_FAST_GPU=1 vllm serve \
+    MiniMaxAI/MiniMax-M2 --trust-remote-code \
+    --tensor-parallel-size 4 \
+    --enable-auto-tool-choice --tool-call-parser minimax_m2 \
+    --reasoning-parser minimax_m2_append_think
+```
+
 8 卡部署命令：
 
 ```bash
@@ -48,8 +58,7 @@ SAFETENSORS_FAST_GPU=1 vllm serve \
     MiniMaxAI/MiniMax-M2 --trust-remote-code \
     --enable_expert_parallel --tensor-parallel-size 8 \
     --enable-auto-tool-choice --tool-call-parser minimax_m2 \
-    --reasoning-parser minimax_m2_append_think \
-    --compilation-config "{\"cudagraph_mode\": \"PIECEWISE\"}"
+    --reasoning-parser minimax_m2_append_think 
 ```
 
 ## 测试部署
@@ -82,6 +91,18 @@ export HF_ENDPOINT=https://hf-mirror.com
 
 该 vLLM 版本过旧，请升级到最新版本。
 
+### torch.AcceleratorError: CUDA error: an illegal memory access was encountered
+在启动参数添加 `--compilation-config "{\"cudagraph_mode\": \"PIECEWISE\"}"` 可以解决。例如：
+
+```bash
+SAFETENSORS_FAST_GPU=1 vllm serve \
+    MiniMaxAI/MiniMax-M2 --trust-remote-code \
+    --enable_expert_parallel --tensor-parallel-size 8 \
+    --enable-auto-tool-choice --tool-call-parser minimax_m2 \
+    --reasoning-parser minimax_m2_append_think \
+    --compilation-config "{\"cudagraph_mode\": \"PIECEWISE\"}"
+```
+
 ## 获取支持
 
 如果在部署 MiniMax 模型过程中遇到任何问题：