I generated an interactive diagram representation of BitNet

[ivanmilevtues]

Hey, all I've always wanted to contribute to open-source projects. But at the very begining I found it a bit too difficult to navigate them and wrap my head around on the main flow and the main "logical" components. That is why I started working on a project to help people like me, who are visual learners. The project generates an interactive diagram starting from the highest level of abstraction and allows you to dive deeper into interesting components.

Here is the diagram:

graph LR
    User_Interface_CLI_Server_["User Interface (CLI/Server)"]
    Environment_Build_System["Environment & Build System"]
    Model_Preparation_Quantization["Model Preparation & Quantization"]
    Inference_Orchestrator_Python_["Inference Orchestrator (Python)"]
    Core_Inference_Engine_C_CUDA_["Core Inference Engine (C++/CUDA)"]
    Tokenizer_Text_Processing["Tokenizer & Text Processing"]
    User_Interface_CLI_Server_ -- "initiates inference request" --> Inference_Orchestrator_Python_
    User_Interface_CLI_Server_ -- "triggers setup/build" --> Environment_Build_System
    Environment_Build_System -- "prepares environment for" --> Model_Preparation_Quantization
    Environment_Build_System -- "compiles & links" --> Core_Inference_Engine_C_CUDA_
    Model_Preparation_Quantization -- "outputs optimized GGUF model" --> Inference_Orchestrator_Python_
    Inference_Orchestrator_Python_ -- "delegates computation to" --> Core_Inference_Engine_C_CUDA_
    Inference_Orchestrator_Python_ -- "sends text/tokens for processing to" --> Tokenizer_Text_Processing
    Core_Inference_Engine_C_CUDA_ -- "returns inference results to" --> Inference_Orchestrator_Python_
    Tokenizer_Text_Processing -- "returns processed text/tokens to" --> Inference_Orchestrator_Python_
    click User_Interface_CLI_Server_ href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/BitNet/User_Interface_CLI_Server_.md" "Details"
    click Environment_Build_System href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/BitNet/Environment_Build_System.md" "Details"
    click Model_Preparation_Quantization href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/BitNet/Model_Preparation_Quantization.md" "Details"
    click Inference_Orchestrator_Python_ href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/BitNet/Inference_Orchestrator_Python_.md" "Details"
    click Core_Inference_Engine_C_CUDA_ href "https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/BitNet/Core_Inference_Engine_C_CUDA_.md" "Details"

Loading

The generation tool is based on Static Analysis and LLMs and is fully open-source: https://github.com/CodeBoarding/CodeBoarding

You can see the full diagram with explanations and related source code here: https://github.com/CodeBoarding/GeneratedOnBoardings/blob/main/mkdocs/on_boarding.md

Any feedback is appreciated! Would love to hear your opinion on diagrams as an exploration tooling and diagram first documentation!