[cookbook] Make container universal (#86)

* Make container universal with `modular/max-full` base image
* Update readme

Author: williamw

genai-cookbook/Dockerfile

@@ -1,25 +1,35 @@
-ARG GPU_TYPE=nvidia
-FROM modular/max-${GPU_TYPE}-base
+# //////////////////////////////////////////////////////////////////////
+# BASE ON UNIVERSAL MAX CONTAINER
+FROM modular/max-full
 
 WORKDIR /app
 
+# //////////////////////////////////////////////////////////////////////
+# UPDATE SYSTEM
+
 RUN apt-get update && apt-get install -y \
     curl \
     build-essential \
     git \
     wget \
     && rm -rf /var/lib/apt/lists/*
 
+# //////////////////////////////////////////////////////////////////////
+# INSTALL NODE & NPM PACKAGES
+
 RUN curl -fsSL https://deb.nodesource.com/setup_18.x | bash - \
-    && apt-get install -y nodejs \
-    && rm -rf /var/lib/apt/lists/*
+&& apt-get install -y nodejs \
+&& rm -rf /var/lib/apt/lists/*
 
 RUN npm install -g pm2 [email protected]
 
 COPY package*.json ./
 
 RUN npm install
 
+# //////////////////////////////////////////////////////////////////////
+# BUILD WEB APP
+
 COPY . /app
 
 RUN npm run build

genai-cookbook/README.md

@@ -175,40 +175,34 @@ To use the cookbook with MAX:
 
 ## Running with Docker
 
-The GenAI Cookbook can be run entirely within a Docker container, including the MAX model server and web application.
+The GenAI Cookbook can be run entirely within a Docker container, including the MAX model server and web application. The container uses the universal MAX image with the nightly build, supporting both NVIDIA and AMD GPUs.
 
 ### Building the Container
 
-The Dockerfile defaults to NVIDIA GPUs:
+Build the universal container image:
 
 ```bash
-docker build --ulimit nofile=65535:65535 -t max-recipes:latest .
-```
-
-Use the `--build-arg` flag to specify AMD:
-
-```bash
-docker build --build-arg GPU_TYPE=amd --ulimit nofile=65535:65535 -t max-recipes:latest .
+docker build --ulimit nofile=65535:65535 -t max-cookbook:latest .
 ```
 
 **Note:** The `--ulimit nofile=65535:65535` flag increases the file descriptor limit, which is needed for Next.js builds.
 
 ### Running the Container
 
-#### With NVIDIA GPUs
+#### NVIDIA GPU
 
 ```bash
 docker run --gpus all \
     -v ~/.cache/huggingface:/root/.cache/huggingface \
     --env "HF_HUB_ENABLE_HF_TRANSFER=1" \
     --env "HF_TOKEN=your-huggingface-token" \
-    --env "MAX_MODEL=google/gemma-3-27b-it" \
+    --env "MAX_MODEL=mistral-community/pixtral-12b" \
     -p 8000:8000 \
     -p 3000:3000 \
-    max-recipes:latest
+    max-cookbook:latest
 ```
 
-#### With AMD GPUs
+#### AMD GPU
 
 ```bash
 docker run \
@@ -219,10 +213,10 @@ docker run \
     -v ~/.cache/huggingface:/root/.cache/huggingface \
     --env "HF_HUB_ENABLE_HF_TRANSFER=1" \
     --env "HF_TOKEN=your-huggingface-token" \
-    --env "MAX_MODEL=google/gemma-3-27b-it" \
+    --env "MAX_MODEL=mistral-community/pixtral-12b" \
     -p 8000:8000 \
     -p 3000:3000 \
-    max-recipes:latest
+    max-cookbook:latest
 ```
 
 **Configuration:**

genai-cookbook/ecosystem.config.js

@@ -2,8 +2,11 @@ module.exports = {
     apps: [
         {
             name: 'max-llm',
-            script: 'max',
-            args: `serve --model ${process.env.MAX_MODEL || 'google/gemma-3-27b-it'} --trust-remote-code`,
+            script: '/bin/bash',
+            args: [
+                '-c',
+                'exec python -m max.entrypoints.pipelines serve --model-path "${MAX_MODEL:-google/gemma-3-27b-it}" --trust-remote-code',
+            ],
             interpreter: 'none',
             autorestart: true,
             watch: false,
@@ -13,13 +16,13 @@ module.exports = {
                 MAX_SERVE_PORT: 8000,
                 NODE_ENV: 'production',
             },
-        },
+    },
         {
             name: 'web-app',
             script: '/bin/bash',
             args: [
                 '-c',
-                'wait-on http-get://0.0.0.0:8000/health -t 600000 -i 2000 && npm start',
+                'wait-on http-get://0.0.0.0:8000/v1/health -t 600000 -i 2000 && npm start',
             ],
             interpreter: 'none',
             autorestart: true,