SGLang Server Doc

docs/interact/sglang.md

@@ -0,0 +1,68 @@
+---
+sidebar_position: 6
+---
+
+# SGLang Server Plugin
+
+The SGLang Server plugin enables multimodal (text + image) and text-only inference using SGLang-compatible models.  
+This plugin is ideal for users who want to experiment with cutting-edge open-source models that support both images and text.
+
+[SGLang](https://docs.sglang.ai/) is a high-performance inference engine optimized for compositional generation and multimodal interactions.
+
+## Getting Started with SGLang Server
+
+### Step 1: Install the SGLang Server Plugin
+
+To get started, first install the SGLang Server plugin:
+
+- Navigate to the **Plugins** section of Transformer Lab.
+- Under **Loader Plugins**, locate and install the plugin named **SGLang Server**.
+
+This enables support for serving models using SGLang's runtime.
+
+### Step 2: Select a Supported Model
+
+The SGLang Server requires models that are explicitly compatible with its runtime.  
+You can find the full list of supported models here:
+
+➡️ [SGLang Supported Models](https://docs.sglang.ai/supported_models/multimodal_language_models.html)
+
+These include both text-only and multimodal models capable of handling image input.
+
+### Step 3: Download a Compatible Model
+
+There are two ways to get an SGLang-compatible model:
+
+**Option A: Download from the Model Zoo**
+
+- Go to the **Model Zoo** tab.
+- Find a compatible model.
+- Click the **Download** button next to the model you wish to use.
+
+**Option B: Use the download bar**
+
+You can use the download bar located at the bottom of the **Model Zoo** tab to download models directly from HuggingFace.
+
+**Option C: Import Your Own**
+
+If you've already prepared a compatible model locally, you can import it using the **Import Local Models** button in the Model Zoo.
+
+### Step 4: Serve the Model
+
+Once the plugin is installed and a model is downloaded:
+
+- Go to the **Foundation** tab.
+- Select the downloaded model.
+- Ensure the **Inference Server** is set to **SGLang Server** (check the menu next to the **Run** button).
+- Click **Run** to start the server.
+
+### Step 5: Interact with Your Model
+
+Once the model is running:
+
+- Go to the **Interact** tab.
+- You can now chat using **text + image** prompts, or send **text-only** completions.
+- The system will route the request through the SGLang runtime based on the model’s capabilities.
+
+<img src={require('./gifs/sglangserver.gif').default} alt="Run with SGLang" width="500" />
+