> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cyberwave.com/llms.txt
> Use this file to discover all available pages before exploring further.

# SmolVLA Training

> Train SmolVLA models using the LeRobot v3 dataset format

## Overview

SmolVLA is a lightweight Vision-Language-Action model optimized for edge deployment. Cyberwave supports fine-tuning SmolVLA models on your custom datasets using the LeRobot v3 dataset format.

<Info>
  SmolVLA models use the **LeRobot v3 dataset format** for training, which differs from the TFDS format used by OpenVLA models. The platform handles format conversion automatically.
</Info>

***

## Model Selection

When starting a new training:

1. Navigate to **AI → Training** in your environment
2. Select **SmolVLA** from the available model architectures
3. The platform will automatically use the LeRobot training pipeline

<Note>
  Only models with `is_trainable: true` appear in the training model selection. SmolVLA is pre-configured as trainable.
</Note>

***

## Dataset Conversion

When you start training with a SmolVLA model, the platform:

1. Joins your episode parquet files into a single dataset
2. Converts the OpenVLA-format parquet to LeRobot v3 format
3. Handles camera role mapping (primary, wrist, secondary)
4. Encodes video frames using AV1 codec (configurable)

The conversion is cached — subsequent trainings with the same dataset and configuration skip the conversion step.

***

## Training Parameters

SmolVLA training supports:

| Parameter     | Description                    | Default   |
| ------------- | ------------------------------ | --------- |
| `fps`         | Target frames per second       | 30        |
| `use_videos`  | Store frames as MP4 videos     | true      |
| `vcodec`      | Video codec                    | libsvtav1 |
| `num_cameras` | Number of camera streams (1-3) | 1         |

***

## Camera Configuration

Camera roles are mapped to LeRobot conventions:

* `primary` → `observation.images.primary`
* `wrist` → `observation.images.wrist`
* `secondary` → `observation.images.secondary`

Configure camera roles in the training wizard or via the API.

***

## Training Workflow

```mermaid theme={null}
flowchart TD
    A[Start Training] --> B{Model Type}
    B -- "SmolVLA" --> C[Convert to LeRobot v3]
    C --> D[Upload Dataset Zip]
    D --> E[Cloud Node Training]
    E --> F[Model Checkpoint]
    B -- "OpenVLA" --> G[Convert to TFDS]
    G --> H[Cloud Node Training]
    H --> F
```

***

## Deployment

After training completes:

1. Deploy the trained model as a controller policy
2. Assign the VLA controller to your robot twin
3. Use natural language prompts to control the robot

<Tip>
  SmolVLA models are optimized for edge inference, making them suitable for real-time robot control with lower latency than larger VLA models.
</Tip>

***

## Related

* [Train VLA Models](/tutorials/train-vla-cyberwave) — Complete training tutorial
* [ML Models Overview](/use-cyberwave/ml-models) — Model capabilities and providers
* [Deploy Models](/use-cyberwave/ml-models/deploy) — Deployment options