CD-29 - Add deploying custom models docs

content/docs/developer/guide/cvm-management.md

@@ -8,9 +8,9 @@ sidebar_position: 5
 
 This guide explains how to start, monitor, and manage Confidential Virtual Machines (CVMs) for Cube AI using the Hardware Abstraction Layer (HAL) with Buildroot-based images and the provided management scripts.
 
-:::info
+<Callout type="info">
 This guide is specifically for CVM deployments using Buildroot images. For public cloud deployments using cloud-init, please refer to the separate cloud deployment documentation instead.
-:::
+</Callout>
 
 ---
 
@@ -32,7 +32,7 @@ Before starting a CVM, ensure the following files exist. These are the default p
 For testing without confidential computing features, use the standard VM mode. This helps confirm that your base image, kernel, and networking all work before adding enclave-specific requirements.
 
 ```bash
-cd /path/to/cube/hal/cloud
+cd /path/to/cube/hal/buildroot
 ./qemu.sh start
 ```
 
@@ -99,7 +99,7 @@ Edit these variables at the top of `cvm-monitor.sh` to customize behavior. Small
 VM_NAME="cube-ai-vm"                        # VM identifier
 CHECK_INTERVAL=30                           # Health check interval (seconds)
 LOG_DIR="/tmp/cube-logs"                    # Log file directory
-QEMU_SCRIPT="/path/to/cube/hal/cloud/qemu.sh"  # Path to QEMU launch script
+QEMU_SCRIPT="/path/to/cube/hal/buildroot/qemu.sh"  # Path to QEMU launch script
 ```
 
 ### Basic Usage

content/docs/developer/guide/hal.md

@@ -8,9 +8,9 @@ sidebar_position: 4
 
 Cube HAL provides the confidential-computing runtime environment for Cube AI using Buildroot to create custom Linux images optimized for confidential computing.
 
-:::info
+<Callout type="info">
 This guide covers HAL image creation using Buildroot. For managing already-built HAL images and CVMs, see the [CVM Management](/docs/developer/guide/cvm-management) guide.
-:::
+</Callout>
 
 ## Overview
 
@@ -92,7 +92,7 @@ Navigate to `Target packages` → `Cube packages` → `cube-agent` to configure
 
 - **ollama** - Ollama package dependency (automatically required when selected as LLM backend)
 - **Install default models** - Current default models included are `tinyllama`, `starcoder2`, and `nomic-bert`)
-- **Custom models to install** - Specify additional models to include (comma-separated)
+- **Custom models to install** - Specify additional models to include (space-separated)
 - **Enable GPU support** - Enable GPU acceleration (requires compatible hardware)
 
 **Important Notes:**

content/docs/developer/guide/index.md

@@ -19,7 +19,7 @@ as private model upload and fine-tuning.
 - **Hardware Abstraction Layer (HAL)**
 - **Cloud-Init** - Ubuntu-based deployment with cloud-init provisioning
 - **CVM Management**
-- **Private Model Upload**
+- **Private Model Upload** - Deploy custom models via HAL build-time config, cloud-init, or runtime upload
 - **Fine-Tuning Models**
 - **Guardrails** - AI safety controls for input validation and output sanitization