# maximevandegar/papers-in-100-lines-of-code architecture

> Architecture diagram of the GitHub repository maximevandegar/papers-in-100-lines-of-code, 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/maximevandegar/papers-in-100-lines-of-code
- Repository: https://github.com/maximevandegar/papers-in-100-lines-of-code

## Overview

This repository is a catalog of compact, mostly independent implementations of research papers—not one integrated application. The user-facing workflow is to choose a paper implementation, provide its expected data or inputs, run its script, and inspect generated predictions, renders, or other results. The implementations span image and 3D reconstruction, neural rendering, diffusion and other generative models, reinforcement learning, and foundational learning methods. The README establishes the catalog’s purpose; sampled source shows some model and rendering internals. Most scripts were not sampled, so their exact runtime wiring is not asserted.

Main entry points: [`A_Pixel_Is_Worth_More_Than_One_3D_Gaussians_in_Single_View_3D_Reconstruction/pixels_to_gaussians.py`](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/A_Pixel_Is_Worth_More_Than_One_3D_Gaussians_in_Single_View_3D_Reconstruction/pixels_to_gaussians.py) (Pixel Gaussians), [`Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/splatter_image.py`](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/splatter_image.py) (Splatter Image), [`Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/gaussian_splatting.py`](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/gaussian_splatting.py) (Gaussian renderer), [`Denoising_Diffusion_Implicit_Models/unet.py`](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Denoising_Diffusion_Implicit_Models/unet.py) (DDIM U-Net)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_reconstruction["Reconstruction and rendering"]
  node_pixel_gaussians["Pixel Gaussians"]
  node_splatter["Splatter Image<br/>[splatter_image.py]"]
  node_gaussian_renderer["Gaussian renderer"]
  node_three_dgs["3D Gaussian Splatting<br/>[3dgs.py]"]
  node_speedy_splat["Speedy Splat<br/>[speedy_splat.py]"]
  node_dvr["Volumetric rendering<br/>[dvr.py]"]
  node_nerf["NeRF<br/>[nerf.py]"]
  node_pixelnerf["PixelNeRF<br/>[pixelnerf.py]"]
end

subgraph group_generation["Generative models"]
  node_gan["GAN<br/>[GANs.py]"]
  node_conditional_gan["Conditional GAN<br/>[cgan.py]"]
  node_latent_diffusion["Latent diffusion model<br/>[model.py]"]
  node_ddpm["Diffusion models"]
  node_ddim_unet["DDIM U-Net<br/>[unet.py]"]
  node_dreambooth["DreamBooth<br/>[dreambooth.py]"]
  node_foldingdiff["Folding diffusion<br/>[foldingdiff.py]"]
  node_realnvp["Real NVP<br/>[real_nvp.py]"]
end

subgraph group_learning["Learning methods"]
  node_adam["Adam optimizer<br/>[adam.py]"]
  node_resnet["ResNet<br/>[resnet.py]"]
  node_maml["MAML<br/>[maml.py]"]
  node_normalization["RMS normalization<br/>[rmsnorm.py]"]
end

subgraph group_control["Reinforcement learning"]
  node_dqn["DQN<br/>[dqn.py]"]
  node_ddqn["Double DQN<br/>[ddqn.py]"]
  node_ppo["PPO<br/>[ppo.py]"]
end

node_user(("Research user"))

node_user -.->|"chooses implementation"| node_pixel_gaussians
node_user -.->|"chooses implementation"| node_splatter
node_user -.->|"chooses implementation"| node_three_dgs
node_user -.->|"chooses implementation"| node_dvr
node_user -.->|"chooses implementation"| node_nerf
node_user -.->|"chooses implementation"| node_gan
node_user -.->|"chooses implementation"| node_latent_diffusion
node_user -.->|"chooses implementation"| node_foldingdiff
node_user -.->|"chooses implementation"| node_realnvp
node_user -.->|"chooses implementation"| node_adam
node_user -.->|"chooses implementation"| node_maml
node_user -.->|"chooses implementation"| node_dqn
node_user -.->|"chooses implementation"| node_ppo
node_pixel_gaussians -->|"imports renderer"| node_gaussian_renderer
node_splatter -->|"uses renderer"| node_gaussian_renderer
node_pixel_gaussians -->|"imports U-Net"| node_ddim_unet
node_splatter -->|"imports U-Net"| node_ddim_unet

click node_pixel_gaussians "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/A_Pixel_Is_Worth_More_Than_One_3D_Gaussians_in_Single_View_3D_Reconstruction/pixels_to_gaussians.py"
click node_splatter "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/splatter_image.py"
click node_gaussian_renderer "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/gaussian_splatting.py"
click node_three_dgs "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/3D_Gaussian_Splatting_for_Real_Time_Radiance_Field_Rendering/3dgs.py"
click node_speedy_splat "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Speedy_Splat_Fast_3D_Gaussian_Splatting_with_Sparse_Pixels_and_Sparse_Primitives/speedy_splat.py"
click node_dvr "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Differentiable_Volumetric_Rendering_Learning_Implicit_3D_Representations_without_3D_Supervision/dvr.py"
click node_nerf "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/NeRF_Representing_Scenes_as_Neural_Radiance_Fields_for_View_Synthesis/nerf.py"
click node_pixelnerf "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/PixelNeRF_Neural_Radiance_Fields_from_One_or_Few_Images/pixelnerf.py"
click node_gan "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Generative_Adversarial_Networks/GANs.py"
click node_conditional_gan "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Conditional_Generative_Adversarial_Nets/cgan.py"
click node_latent_diffusion "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/High_Resolution_Image_Synthesis_with_Latent_Diffusion_Models/model.py"
click node_ddpm "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Denoising_Diffusion_Probabilistic_Models/diffusion_models.py"
click node_ddim_unet "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Denoising_Diffusion_Implicit_Models/unet.py"
click node_dreambooth "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/DreamBooth_Fine_Tuning_Text_to_Image_Diffusion_Models_for_Subject_Driven_Generation/dreambooth.py"
click node_foldingdiff "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Protein_Structure_Generation_via_Folding_Diffusion/foldingdiff.py"
click node_realnvp "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Density_Estimation_Using_Real_NVP/real_nvp.py"
click node_adam "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Adam_a_Method_For_Stochastic_Optimization/adam.py"
click node_resnet "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Deep_Residual_Learning_for_Image_Recognition/resnet.py"
click node_maml "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Model_Agnostic_Meta_Learning_for_Fast_Adaptation_of_Deep_Networks/maml.py"
click node_normalization "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Root_Mean_Square_Layer_Normalization/rmsnorm.py"
click node_dqn "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Playing_Atari_with_Deep_Reinforcement_Learning/dqn.py"
click node_ddqn "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Deep_Reinforcement_Learning_with_Double_Q_learning/ddqn.py"
click node_ppo "https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Proximal_Policy_Optimization_Algorithms/ppo.py"

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_pixel_gaussians,node_splatter,node_gaussian_renderer,node_three_dgs,node_speedy_splat,node_dvr,node_nerf,node_pixelnerf,node_user toneBlue
class node_gan,node_conditional_gan,node_latent_diffusion,node_ddpm,node_ddim_unet,node_dreambooth,node_foldingdiff,node_realnvp toneAmber
class node_adam,node_resnet,node_maml,node_normalization toneMint
class node_dqn,node_ddqn,node_ppo toneRose
```

## Components

### Reconstruction and rendering

- [Pixel Gaussians](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/A_Pixel_Is_Worth_More_Than_One_3D_Gaussians_in_Single_View_3D_Reconstruction/pixels_to_gaussians.py): `A_Pixel_Is_Worth_More_Than_One_3D_Gaussians_in_Single_View_3D_Reconstruction/pixels_to_gaussians.py`
- [Splatter Image](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/splatter_image.py): `Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/splatter_image.py`
- [Gaussian renderer](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/gaussian_splatting.py): `Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/gaussian_splatting.py`
- [3D Gaussian Splatting](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/3D_Gaussian_Splatting_for_Real_Time_Radiance_Field_Rendering/3dgs.py): `3D_Gaussian_Splatting_for_Real_Time_Radiance_Field_Rendering/3dgs.py`
- [Speedy Splat](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Speedy_Splat_Fast_3D_Gaussian_Splatting_with_Sparse_Pixels_and_Sparse_Primitives/speedy_splat.py): `Speedy_Splat_Fast_3D_Gaussian_Splatting_with_Sparse_Pixels_and_Sparse_Primitives/speedy_splat.py`
- [Volumetric rendering](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Differentiable_Volumetric_Rendering_Learning_Implicit_3D_Representations_without_3D_Supervision/dvr.py): `Differentiable_Volumetric_Rendering_Learning_Implicit_3D_Representations_without_3D_Supervision/dvr.py`
- [NeRF](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/NeRF_Representing_Scenes_as_Neural_Radiance_Fields_for_View_Synthesis/nerf.py): `NeRF_Representing_Scenes_as_Neural_Radiance_Fields_for_View_Synthesis/nerf.py`
- [PixelNeRF](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/PixelNeRF_Neural_Radiance_Fields_from_One_or_Few_Images/pixelnerf.py): `PixelNeRF_Neural_Radiance_Fields_from_One_or_Few_Images/pixelnerf.py`

### Generative models

- [GAN](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Generative_Adversarial_Networks/GANs.py): `Generative_Adversarial_Networks/GANs.py`
- [Conditional GAN](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Conditional_Generative_Adversarial_Nets/cgan.py): `Conditional_Generative_Adversarial_Nets/cgan.py`
- [Latent diffusion model](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/High_Resolution_Image_Synthesis_with_Latent_Diffusion_Models/model.py): `High_Resolution_Image_Synthesis_with_Latent_Diffusion_Models/model.py`
- [Diffusion models](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Denoising_Diffusion_Probabilistic_Models/diffusion_models.py): `Denoising_Diffusion_Probabilistic_Models/diffusion_models.py`
- [DDIM U-Net](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Denoising_Diffusion_Implicit_Models/unet.py): `Denoising_Diffusion_Implicit_Models/unet.py`
- [DreamBooth](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/DreamBooth_Fine_Tuning_Text_to_Image_Diffusion_Models_for_Subject_Driven_Generation/dreambooth.py): `DreamBooth_Fine_Tuning_Text_to_Image_Diffusion_Models_for_Subject_Driven_Generation/dreambooth.py`
- [Folding diffusion](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Protein_Structure_Generation_via_Folding_Diffusion/foldingdiff.py): `Protein_Structure_Generation_via_Folding_Diffusion/foldingdiff.py`
- [Real NVP](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Density_Estimation_Using_Real_NVP/real_nvp.py): `Density_Estimation_Using_Real_NVP/real_nvp.py`

### Learning methods

- [Adam optimizer](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Adam_a_Method_For_Stochastic_Optimization/adam.py): `Adam_a_Method_For_Stochastic_Optimization/adam.py`
- [ResNet](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Deep_Residual_Learning_for_Image_Recognition/resnet.py): `Deep_Residual_Learning_for_Image_Recognition/resnet.py`
- [MAML](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Model_Agnostic_Meta_Learning_for_Fast_Adaptation_of_Deep_Networks/maml.py): `Model_Agnostic_Meta_Learning_for_Fast_Adaptation_of_Deep_Networks/maml.py`
- [RMS normalization](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Root_Mean_Square_Layer_Normalization/rmsnorm.py): `Root_Mean_Square_Layer_Normalization/rmsnorm.py`

### Reinforcement learning

- [DQN](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Playing_Atari_with_Deep_Reinforcement_Learning/dqn.py): `Playing_Atari_with_Deep_Reinforcement_Learning/dqn.py`
- [Double DQN](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Deep_Reinforcement_Learning_with_Double_Q_learning/ddqn.py): `Deep_Reinforcement_Learning_with_Double_Q_learning/ddqn.py`
- [PPO](https://github.com/maximevandegar/papers-in-100-lines-of-code/blob/main/Proximal_Policy_Optimization_Algorithms/ppo.py): `Proximal_Policy_Optimization_Algorithms/ppo.py`

### Other components

- **Research user**

## Connections

- Research user → Pixel Gaussians: chooses implementation
- Research user → Splatter Image: chooses implementation
- Research user → 3D Gaussian Splatting: chooses implementation
- Research user → Volumetric rendering: chooses implementation
- Research user → NeRF: chooses implementation
- Research user → GAN: chooses implementation
- Research user → Latent diffusion model: chooses implementation
- Research user → Folding diffusion: chooses implementation
- Research user → Real NVP: chooses implementation
- Research user → Adam optimizer: chooses implementation
- Research user → MAML: chooses implementation
- Research user → DQN: chooses implementation
- Research user → PPO: chooses implementation
- Pixel Gaussians → Gaussian renderer: imports renderer (evidence: `A_Pixel_Is_Worth_More_Than_One_3D_Gaussians_in_Single_View_3D_Reconstruction/pixels_to_gaussians.py`)
- Splatter Image → Gaussian renderer: uses renderer (evidence: `Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/splatter_image.py`)
- Pixel Gaussians → DDIM U-Net: imports U-Net (evidence: `A_Pixel_Is_Worth_More_Than_One_3D_Gaussians_in_Single_View_3D_Reconstruction/pixels_to_gaussians.py`)
- Splatter Image → DDIM U-Net: imports U-Net (evidence: `Splatter_Image_Ultra_Fast_Single_View_3D_Reconstruction/splatter_image.py`)