# inclusionai/ming-image architecture

> Architecture diagram of the GitHub repository inclusionai/ming-image, 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/inclusionai/ming-image
- Repository: https://github.com/inclusionai/ming-image

## Overview

Ming Image is a model-inference repository for visual-design generation and editable-layer decomposition. The principal workflow accepts a prompt or input image through the inference CLI, loads a checkpoint under a capability contract, and produces a generated design or transparent editable layers. The repository also includes multimodal input processing and the image-generation diffusion stack. The README documents both task paths, but sampled excerpts do not show the CLI’s full dispatch or the layer-decomposition implementation, so those links are omitted rather than inferred.

Main entry points: [`infer.py`](https://github.com/inclusionai/ming-image/blob/main/infer.py) (Inference CLI), [`modeling_bailingmm2.py`](https://github.com/inclusionai/ming-image/blob/main/modeling_bailingmm2.py) (Multimodal model), [`diffusion/transformer.py`](https://github.com/inclusionai/ming-image/blob/main/diffusion/transformer.py) (Diffusion transformer), [`modeling_bailing_moe_v2.py`](https://github.com/inclusionai/ming-image/blob/main/modeling_bailing_moe_v2.py) (MoE language model)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_entry["Inference entry"]
  node_cli["Inference CLI<br/>[infer.py]"]
  node_device["Device placement<br/>[mllm_device_map.py]"]
end

subgraph group_multimodal["Multimodal model"]
  node_mmmodel["Multimodal model"]
  node_moe["MoE language model"]
  node_vision["Vision encoder<br/>[qwen2_5_vit.py]"]
  node_configmm["Multimodal config"]
  node_configmoe["MoE config"]
  node_modelutils["Model utilities<br/>[modeling_utils.py]"]
end

subgraph group_diffusion["Image generation"]
  node_generator["Image generator<br/>[generator.py]"]
  node_pipeline["Denoising pipeline<br/>[pipeline.py]"]
  node_transformer["Diffusion transformer<br/>[transformer.py]"]
  node_vae["Image autoencoder"]
  node_padding["Sequence padding<br/>[padding.py]"]
end

subgraph group_contracts["Input and contracts"]
  node_profile["Checkpoint contract"]
  node_processor["Multimodal processor"]
  node_imageproc["Image preprocessing"]
  node_utils["Media utilities<br/>[bailingmm_utils.py]"]
  node_tokenizer["Bailing tokenizer"]
end

node_user(("User"))
node_design["Generated design"]
node_layers["Editable layers"]

node_user -->|"submits task"| node_cli
node_cli -->|"validates checkpoint"| node_profile
node_mmmodel -->|"constructs"| node_moe
node_mmmodel -->|"constructs"| node_vision
node_mmmodel -->|"constructs"| node_generator
node_transformer -->|"uses"| node_padding
node_cli -.->|"writes result"| node_design
node_cli -.->|"writes result"| node_layers

click node_cli "https://github.com/inclusionai/ming-image/blob/main/infer.py"
click node_profile "https://github.com/inclusionai/ming-image/blob/main/inference_profile.py"
click node_device "https://github.com/inclusionai/ming-image/blob/main/mllm_device_map.py"
click node_processor "https://github.com/inclusionai/ming-image/blob/main/processing_bailingmm2.py"
click node_imageproc "https://github.com/inclusionai/ming-image/blob/main/image_processing_bailingmm2.py"
click node_utils "https://github.com/inclusionai/ming-image/blob/main/bailingmm_utils.py"
click node_tokenizer "https://github.com/inclusionai/ming-image/blob/main/tokenization_bailing.py"
click node_mmmodel "https://github.com/inclusionai/ming-image/blob/main/modeling_bailingmm2.py"
click node_moe "https://github.com/inclusionai/ming-image/blob/main/modeling_bailing_moe_v2.py"
click node_vision "https://github.com/inclusionai/ming-image/blob/main/qwen2_5_vit.py"
click node_configmm "https://github.com/inclusionai/ming-image/blob/main/configuration_bailingmm2.py"
click node_configmoe "https://github.com/inclusionai/ming-image/blob/main/configuration_bailing_moe_v2.py"
click node_modelutils "https://github.com/inclusionai/ming-image/blob/main/modeling_utils.py"
click node_generator "https://github.com/inclusionai/ming-image/blob/main/diffusion/generator.py"
click node_pipeline "https://github.com/inclusionai/ming-image/blob/main/diffusion/pipeline.py"
click node_transformer "https://github.com/inclusionai/ming-image/blob/main/diffusion/transformer.py"
click node_vae "https://github.com/inclusionai/ming-image/blob/main/diffusion/autoencoder_kl_qwenimage.py"
click node_padding "https://github.com/inclusionai/ming-image/blob/main/diffusion/padding.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_cli,node_device,node_user toneBlue
class node_mmmodel,node_moe,node_vision,node_configmm,node_configmoe,node_modelutils toneAmber
class node_generator,node_pipeline,node_transformer,node_vae,node_padding toneMint
class node_profile,node_processor,node_imageproc,node_utils,node_tokenizer toneRose
class node_design,node_layers toneIndigo
```

## Components

### Inference entry

- [Inference CLI](https://github.com/inclusionai/ming-image/blob/main/infer.py): `infer.py`
- [Device placement](https://github.com/inclusionai/ming-image/blob/main/mllm_device_map.py): `mllm_device_map.py`

### Multimodal model

- [Multimodal model](https://github.com/inclusionai/ming-image/blob/main/modeling_bailingmm2.py): `modeling_bailingmm2.py`
- [MoE language model](https://github.com/inclusionai/ming-image/blob/main/modeling_bailing_moe_v2.py): `modeling_bailing_moe_v2.py`
- [Vision encoder](https://github.com/inclusionai/ming-image/blob/main/qwen2_5_vit.py): `qwen2_5_vit.py`
- [Multimodal config](https://github.com/inclusionai/ming-image/blob/main/configuration_bailingmm2.py): `configuration_bailingmm2.py`
- [MoE config](https://github.com/inclusionai/ming-image/blob/main/configuration_bailing_moe_v2.py): `configuration_bailing_moe_v2.py`
- [Model utilities](https://github.com/inclusionai/ming-image/blob/main/modeling_utils.py): `modeling_utils.py`

### Image generation

- [Image generator](https://github.com/inclusionai/ming-image/blob/main/diffusion/generator.py): `diffusion/generator.py`
- [Denoising pipeline](https://github.com/inclusionai/ming-image/blob/main/diffusion/pipeline.py): `diffusion/pipeline.py`
- [Diffusion transformer](https://github.com/inclusionai/ming-image/blob/main/diffusion/transformer.py): `diffusion/transformer.py`
- [Image autoencoder](https://github.com/inclusionai/ming-image/blob/main/diffusion/autoencoder_kl_qwenimage.py): `diffusion/autoencoder_kl_qwenimage.py`
- [Sequence padding](https://github.com/inclusionai/ming-image/blob/main/diffusion/padding.py): `diffusion/padding.py`

### Input and contracts

- [Checkpoint contract](https://github.com/inclusionai/ming-image/blob/main/inference_profile.py): `inference_profile.py`
- [Multimodal processor](https://github.com/inclusionai/ming-image/blob/main/processing_bailingmm2.py): `processing_bailingmm2.py`
- [Image preprocessing](https://github.com/inclusionai/ming-image/blob/main/image_processing_bailingmm2.py): `image_processing_bailingmm2.py`
- [Media utilities](https://github.com/inclusionai/ming-image/blob/main/bailingmm_utils.py): `bailingmm_utils.py`
- [Bailing tokenizer](https://github.com/inclusionai/ming-image/blob/main/tokenization_bailing.py): `tokenization_bailing.py`

### Other components

- **User**
- **Generated design**
- **Editable layers**

## Connections

- User → Inference CLI: submits task (evidence: `README.md`)
- Inference CLI → Checkpoint contract: validates checkpoint (evidence: `README.md`)
- Multimodal model → MoE language model: constructs (evidence: `modeling_bailingmm2.py`)
- Multimodal model → Vision encoder: constructs (evidence: `modeling_bailingmm2.py`)
- Multimodal model → Image generator: constructs (evidence: `modeling_bailingmm2.py`)
- Diffusion transformer → Sequence padding: uses (evidence: `diffusion/transformer.py`)
- Inference CLI → Generated design: writes result
- Inference CLI → Editable layers: writes result
