# git03-khushi/jit-cifar10-image-generation architecture

> Architecture diagram of the GitHub repository git03-khushi/jit-cifar10-image-generation, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-12). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/git03-khushi/jit-cifar10-image-generation
- Repository: https://github.com/git03-khushi/jit-cifar10-image-generation

## Overview

This repository is a notebook-based, reduced-scale JiT-inspired diffusion experiment for CIFAR-10. Its documented workflow loads and preprocesses data, trains a patch-based Transformer denoiser under multiple patch-size and time-embedding configurations, then evaluates and visualizes results and saves checkpoints. The notebook itself is the only implementation file in the tree; component boundaries and edges below reflect README descriptions, not inspected notebook cells.

Main entry points: [`JiT Project.ipynb`](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb) (Model training)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Data preparation"]
  node_preprocess["Image preprocessing<br/>[JiT Project.ipynb]"]
  node_noisy_input["Noisy input<br/>[JiT Project.ipynb]"]
end

subgraph group_model["Denoising model"]
  node_patch_model["Patch Transformer<br/>[JiT Project.ipynb]"]
  node_time_embedding["Time embedding<br/>[JiT Project.ipynb]"]
  node_clean_prediction["Clean-image prediction<br/>[JiT Project.ipynb]"]
  node_velocity_objective["Velocity MSE objective<br/>[JiT Project.ipynb]"]
end

subgraph group_experiments["Training experiments"]
  node_patch4["Patch-4 run<br/>[JiT Project.ipynb]"]
  node_patch8["Patch-8 run<br/>[JiT Project.ipynb]"]
  node_ablation["Time ablation<br/>[JiT Project.ipynb]"]
  node_training["Model training<br/>[JiT Project.ipynb]"]
end

subgraph group_results["Evaluation outputs"]
  node_test_loss["Test-loss evaluation<br/>[JiT Project.ipynb]"]
  node_loss_viz["Loss visualization<br/>[JiT Project.ipynb]"]
  node_sample_viz["Sample and denoising views<br/>[JiT Project.ipynb]"]
  node_checkpoint["Checkpoint saving<br/>[JiT Project.ipynb]"]
end

node_researcher(("Researcher"))
node_cifar[("CIFAR-10")]

node_researcher -->|"runs notebook"| node_preprocess
node_cifar -->|"loads data"| node_preprocess
node_preprocess -->|"prepares images"| node_noisy_input
node_noisy_input -->|"feeds denoiser"| node_patch_model
node_time_embedding -->|"conditions model"| node_patch_model
node_patch_model -->|"predicts clean image"| node_clean_prediction
node_clean_prediction -->|"derives prediction"| node_velocity_objective
node_patch4 -->|"trains configuration"| node_training
node_patch8 -->|"trains configuration"| node_training
node_ablation -->|"trains configuration"| node_training
node_training -->|"minimizes loss"| node_velocity_objective
node_training -->|"evaluates model"| node_test_loss
node_training -->|"visualizes loss"| node_loss_viz
node_training -->|"generates visualizations"| node_sample_viz
node_training -->|"saves weights"| node_checkpoint

click node_preprocess "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_noisy_input "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_patch_model "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_time_embedding "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_clean_prediction "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_velocity_objective "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_patch4 "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_patch8 "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_ablation "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_training "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_test_loss "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_loss_viz "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_sample_viz "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"
click node_checkpoint "https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb"

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_preprocess,node_noisy_input toneBlue
class node_patch_model,node_time_embedding,node_clean_prediction,node_velocity_objective,node_cifar toneAmber
class node_patch4,node_patch8,node_ablation,node_training toneMint
class node_test_loss,node_loss_viz,node_sample_viz,node_checkpoint toneRose
class node_researcher toneIndigo
```

## Components

### Data preparation

- [Image preprocessing](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Noisy input](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`

### Denoising model

- [Patch Transformer](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Time embedding](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Clean-image prediction](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Velocity MSE objective](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`

### Training experiments

- [Patch-4 run](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Patch-8 run](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Time ablation](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Model training](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`

### Evaluation outputs

- [Test-loss evaluation](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Loss visualization](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Sample and denoising views](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`
- [Checkpoint saving](https://github.com/git03-khushi/jit-cifar10-image-generation/blob/main/JiT%20Project.ipynb): `JiT Project.ipynb`

### Other components

- **Researcher**
- **CIFAR-10**

## Connections

- Researcher → Image preprocessing: runs notebook (evidence: `README.md`)
- CIFAR-10 → Image preprocessing: loads data (evidence: `README.md`)
- Image preprocessing → Noisy input: prepares images (evidence: `README.md`)
- Noisy input → Patch Transformer: feeds denoiser (evidence: `README.md`)
- Time embedding → Patch Transformer: conditions model (evidence: `README.md`)
- Patch Transformer → Clean-image prediction: predicts clean image (evidence: `README.md`)
- Clean-image prediction → Velocity MSE objective: derives prediction (evidence: `README.md`)
- Patch-4 run → Model training: trains configuration (evidence: `README.md`)
- Patch-8 run → Model training: trains configuration (evidence: `README.md`)
- Time ablation → Model training: trains configuration (evidence: `README.md`)
- Model training → Velocity MSE objective: minimizes loss (evidence: `README.md`)
- Model training → Test-loss evaluation: evaluates model (evidence: `README.md`)
- Model training → Loss visualization: visualizes loss (evidence: `README.md`)
- Model training → Sample and denoising views: generates visualizations (evidence: `README.md`)
- Model training → Checkpoint saving: saves weights (evidence: `README.md`)