# llm-jp/awesome-japanese-llm architecture

> Architecture diagram of the GitHub repository llm-jp/awesome-japanese-llm, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2025-04-10). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/llm-jp/awesome-japanese-llm
- Repository: https://github.com/llm-jp/awesome-japanese-llm

## Overview

##### Project’s Overall Purpose and Type
- This “awesome” repository serves as a centralized index for Japanese LLM research and resources. It is both a directory and a reference documentation hub.
- Its purpose is to catalog various large language models (LLMs) (generative, autoencoding, embeddings, multimodal, speech, etc.), evaluation benchmarks (general, domain‐specific, logical/safety, etc.), and pointers to original research (reference papers).

##### Major Components
A. Resource / Model Catalog:
– This includes sections such as:
- Generative (text generation) models
- Models for autoencoding / input text processing (e.g., BERT, RoBERTa, ELECTRA, etc.)
- Embeddings – single and multi-representation bi-encoders and cross-encoders
- Multimodal models which further subdivide into:
- Vision-Language models for image+text generation or VQA
- Text-to-image generation models (including domain-specific ones)
- Text-to-video generation models
- Speech / Audio models (ASR and other speech-language models)
B. Evaluation and Benchmark Suite:
– A separate “layer” in the documentation shows benchmarks and evaluation datasets:
- Hybrid/General language evaluation metrics (Japanese MT-bench, JMMLU, etc.)
- Domain-specific benchmarks (financial, medical, logical reasoning, controllability)
- Embedding benchmarks, VLM evaluation tools, etc.
C. Supporting Documentation and References:
– A part of the repository is devoted to listing primary research papers and contributing guidelines.
– There is also a section for citation information and contributor details.

##### Relationships and Dependencies
- The “catalog” component feeds into the “evaluation” layer: several benchmarks reference models described in the catalog.
- There are cross-references between each category – for example, a given model might appear in multiple sections (e.g., Japanese LLMs appear both as generative models and as the basis for multimodal systems).
- The evaluation benchmarks also refer to external data sources (e.g., translation of benchmarks, curated datasets like JA-VG-VQA, etc.).
- The “reference paper” section supports and documents the methodology behind many entries (and may conceptually “overlay” many of the above components).

## Diagram (Mermaid)

```mermaid
graph TD
    %% Top Level: Central Repository Hub
    A("Central Repository Hub"):::hub

    %% Resource Catalog Subgraph
    subgraph "Resource Catalog"
        CA("Catalog Index"):::catalog
        B("Generative Models (e.g., GPT-Ja)"):::catalog
        C("Autoencoding/Text Models (e.g., ModernBERT-Ja)"):::catalog
        D("Embedding Models"):::catalog
        E("Multimodal Models"):::multimodal
            subgraph "Multimodal Subsections"
                E1("Vision-Language"):::multimodal
                E2("Text-to-Image"):::multimodal
                E3("Text-to-Video"):::multimodal
            end
        F("Speech Models"):::catalog
        G("API-based Offerings"):::catalog
    end

    %% Evaluation & Benchmarks Subgraph
    subgraph "Evaluation & Benchmarks"
        H("General Language Evaluation (MT-bench, JMMLU)"):::benchmark
        I("Domain-specific Benchmarks"):::benchmark
        J("Embedding Retrieval Benchmarks"):::benchmark
        K("VLM Benchmarks"):::benchmark
    end

    %% Supporting Documentation and References Subgraph
    subgraph "Supporting Documentation and References"
        L("Citation Details"):::doc
        M("Contributor Guidelines"):::doc
        N("Model-related References"):::doc
        O("Training & Research References"):::doc
    end

    %% Continuous Integration / Update Mechanism
    P("Continuous Integration & Update Mechanism"):::ci

    %% Connections from Central Repository Hub to other components
    A -->|"hosts"| CA
    CA -->|"includes"| B
    CA -->|"includes"| C
    CA -->|"includes"| D
    CA -->|"includes"| E
    CA -->|"includes"| F
    CA -->|"includes"| G

    A -->|"supplies"| H
    A -->|"supplies"| I
    A -->|"supplies"| J
    A -->|"supplies"| K

    A -->|"documents"| L
    A -->|"documents"| M
    A -->|"documents"| N
    A -->|"documents"| O

    P -->|"monitors"| A

    %% Additional flows from specific Resource Catalog nodes into Evaluation & Benchmarks
    B -->|"input"| H
    C -->|"input"| H
    D -->|"input"| J
    E -->|"input"| K
    F -->|"input"| I
    G -->|"input"| I

    %% Styles
    classDef hub fill:#FFE4B5,stroke:#333,stroke-width:2px;
    classDef catalog fill:#ADD8E6,stroke:#333,stroke-width:1px;
    classDef multimodal fill:#B0E0E6,stroke:#333,stroke-width:1px;
    classDef benchmark fill:#98FB98,stroke:#333,stroke-width:1px;
    classDef doc fill:#FFDAB9,stroke:#333,stroke-width:1px;
    classDef ci fill:#FFB6C1,stroke:#333,stroke-width:1px;

    class A hub;
    class CA,B,C,D,F,G catalog;
    class E,E1,E2,E3 multimodal;
    class H,I,J,K benchmark;
    class L,M,N,O doc;
    class P ci;

    %% Click Events
    click A "https://github.com/llm-jp/awesome-japanese-llm/blob/main/README.md"
    click CA "https://github.com/llm-jp/awesome-japanese-llm/blob/main/README.md"
    click L "https://github.com/llm-jp/awesome-japanese-llm/blob/main/CITATION.cff"
    click M "https://github.com/llm-jp/awesome-japanese-llm/blob/main/CONTRIBUTING.md"
    click N "https://github.com/llm-jp/awesome-japanese-llm/blob/main/parts/references_model.md"
    click O "https://github.com/llm-jp/awesome-japanese-llm/blob/main/parts/references_training.md"
    click P "https://github.com/llm-jp/awesome-japanese-llm/tree/main/.github/workflows"
```
