# huytai1o2/smolvla architecture

> Architecture diagram of the GitHub repository huytai1o2/smolvla, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-10-01). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/huytai1o2/smolvla
- Repository: https://github.com/huytai1o2/smolvla

## Overview

This repository packages a reproducible LIBERO orange-juice robot rollout: install the evaluation environment and fine-tuned policy, then run a closed-loop episode and inspect its success metrics and video. The documented workflow spans setup, policy/simulation interaction, and result generation. Relationships below are based on the README only; no script source excerpts were available, so implementation details beyond those documented are omitted.

Main entry points: [`scripts/install.sh`](https://github.com/huytai1o2/smolvla/blob/main/scripts/install.sh) (Environment Installer), [`scripts/run_orange_juice.sh`](https://github.com/huytai1o2/smolvla/blob/main/scripts/run_orange_juice.sh) (Episode Runner)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_setup["Local Setup"]
  node_installer["Environment Installer<br/>[install.sh]"]
  node_lerobot["LeRobot Runtime"]
  node_policy["Fine-tuned Policy"]
  node_libero_assets["LIBERO Assets"]
end

subgraph group_rollout["Robot Rollout"]
  node_runner["Episode Runner"]
  node_evaluator["LIBERO Evaluator"]
  node_simulation["Robot Simulation"]
  node_task["Object Task 9"]
end

subgraph group_results["Evaluation Results"]
  node_metrics["Episode Metrics"]
  node_video["Raw Rollout Video"]
end

node_user(("User"))
node_huggingface["Hugging Face"]

node_user -->|"runs setup"| node_installer
node_installer -->|"installs"| node_lerobot
node_installer -->|"downloads"| node_policy
node_installer -->|"downloads"| node_libero_assets
node_huggingface -->|"hosts checkpoint"| node_policy
node_user -->|"starts episode"| node_runner
node_runner -->|"runs evaluation"| node_evaluator
node_evaluator -->|"evaluates"| node_task
node_evaluator -->|"runs rollout"| node_simulation
node_evaluator -->|"queries actions"| node_policy
node_simulation -->|"provides observations"| node_policy
node_evaluator -->|"writes results"| node_metrics
node_evaluator -->|"generates video"| node_video
node_user -->|"inspects success"| node_metrics
node_user -->|"views rollout"| node_video

click node_installer "https://github.com/huytai1o2/smolvla/blob/main/scripts/install.sh"
click node_runner "https://github.com/huytai1o2/smolvla/blob/main/scripts/run_orange_juice.sh"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_installer,node_lerobot,node_policy,node_libero_assets,node_user toneBlue
class node_runner,node_evaluator,node_simulation,node_task toneAmber
class node_metrics,node_video toneMint
class node_huggingface toneIndigo
```

## Components

### Local Setup

- [Environment Installer](https://github.com/huytai1o2/smolvla/blob/main/scripts/install.sh): `scripts/install.sh`
- **LeRobot Runtime**
- **Fine-tuned Policy**
- **LIBERO Assets**

### Robot Rollout

- [Episode Runner](https://github.com/huytai1o2/smolvla/blob/main/scripts/run_orange_juice.sh): `scripts/run_orange_juice.sh`
- **LIBERO Evaluator**
- **Robot Simulation**
- **Object Task 9**

### Evaluation Results

- **Episode Metrics**
- **Raw Rollout Video**

### Other components

- **User**
- **Hugging Face**

## Connections

- User → Environment Installer: runs setup (evidence: `README.md`)
- Environment Installer → LeRobot Runtime: installs (evidence: `README.md`)
- Environment Installer → Fine-tuned Policy: downloads (evidence: `README.md`)
- Environment Installer → LIBERO Assets: downloads (evidence: `README.md`)
- Hugging Face → Fine-tuned Policy: hosts checkpoint (evidence: `README.md`)
- User → Episode Runner: starts episode (evidence: `README.md`)
- Episode Runner → LIBERO Evaluator: runs evaluation (evidence: `README.md`)
- LIBERO Evaluator → Object Task 9: evaluates (evidence: `README.md`)
- LIBERO Evaluator → Robot Simulation: runs rollout (evidence: `README.md`)
- LIBERO Evaluator → Fine-tuned Policy: queries actions (evidence: `README.md`)
- Robot Simulation → Fine-tuned Policy: provides observations (evidence: `README.md`)
- LIBERO Evaluator → Episode Metrics: writes results (evidence: `README.md`)
- LIBERO Evaluator → Raw Rollout Video: generates video (evidence: `README.md`)
- User → Episode Metrics: inspects success (evidence: `README.md`)
- User → Raw Rollout Video: views rollout (evidence: `README.md`)
