# konpep-dev/aether-rwkv- architecture

> Architecture diagram of the GitHub repository konpep-dev/aether-rwkv-, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-22). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/konpep-dev/aether-rwkv-
- Repository: https://github.com/konpep-dev/aether-rwkv-

## Overview

Aether is a bilingual RWKV language-model project. Its product workflow runs from synthetic-corpus creation and tokenization through model training, then interactive prompt encoding, recurrent token generation, and decoded chat output. The graph preserves the documented dataset, tokenizer, training, inference, and RWKV capabilities. Some subsystem links rely on README documentation because the excerpts are partial; unsampled implementation details are not inferred.

Main entry points: [`train.py`](https://github.com/konpep-dev/aether-rwkv-/blob/main/train.py) (Training loop), [`inference.py`](https://github.com/konpep-dev/aether-rwkv-/blob/main/inference.py) (Chat inference), [`model.py`](https://github.com/konpep-dev/aether-rwkv-/blob/main/model.py) (RWKV model), [`tokenizer.py`](https://github.com/konpep-dev/aether-rwkv-/blob/main/tokenizer.py) (Byte-BPE tokenizer)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Data preparation"]
  node_dataset_generator["Corpus generator"]
  node_corpus["Synthetic corpus"]
  node_tokenizer["Byte-BPE tokenizer<br/>[tokenizer.py]"]
end

subgraph group_learning["Model training"]
  node_trainer["Training loop<br/>[train.py]"]
  node_checkpoint["Model checkpoint"]
end

subgraph group_runtime["Chat inference"]
  node_inference["Chat inference<br/>[inference.py]"]
end

subgraph group_architecture["RWKV engine"]
  node_rwkv["RWKV model<br/>[model.py]"]
  node_time_mix["Time-mix recurrence<br/>[model.py]"]
  node_channel_mix["Channel-mix layer<br/>[model.py]"]
  node_wkv_kernel["WKV acceleration<br/>[wkv_cuda_kernel.py]"]
end

node_user(("Chat user"))

node_dataset_generator -->|"generates"| node_corpus
node_corpus -.->|"training data"| node_trainer
node_trainer -->|"encodes samples"| node_tokenizer
node_trainer -->|"trains"| node_rwkv
node_trainer -->|"saves weights"| node_checkpoint
node_checkpoint -.->|"loads weights"| node_inference
node_user -->|"submits prompts"| node_inference
node_inference -->|"encodes and decodes"| node_tokenizer
node_inference -->|"generates tokens"| node_rwkv
node_rwkv -->|"applies"| node_time_mix
node_rwkv -->|"applies"| node_channel_mix
node_time_mix -->|"dispatches WKV"| node_wkv_kernel

click node_dataset_generator "https://github.com/konpep-dev/aether-rwkv-/blob/main/generate_dataset.py"
click node_tokenizer "https://github.com/konpep-dev/aether-rwkv-/blob/main/tokenizer.py"
click node_trainer "https://github.com/konpep-dev/aether-rwkv-/blob/main/train.py"
click node_inference "https://github.com/konpep-dev/aether-rwkv-/blob/main/inference.py"
click node_rwkv "https://github.com/konpep-dev/aether-rwkv-/blob/main/model.py"
click node_time_mix "https://github.com/konpep-dev/aether-rwkv-/blob/main/model.py"
click node_channel_mix "https://github.com/konpep-dev/aether-rwkv-/blob/main/model.py"
click node_wkv_kernel "https://github.com/konpep-dev/aether-rwkv-/blob/main/wkv_cuda_kernel.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_dataset_generator,node_corpus,node_tokenizer,node_user toneBlue
class node_trainer,node_checkpoint toneAmber
class node_inference toneMint
class node_rwkv,node_time_mix,node_channel_mix,node_wkv_kernel toneRose
```

## Components

### Data preparation

- [Corpus generator](https://github.com/konpep-dev/aether-rwkv-/blob/main/generate_dataset.py): `generate_dataset.py`
- **Synthetic corpus**
- [Byte-BPE tokenizer](https://github.com/konpep-dev/aether-rwkv-/blob/main/tokenizer.py): `tokenizer.py`

### Model training

- [Training loop](https://github.com/konpep-dev/aether-rwkv-/blob/main/train.py): `train.py`
- **Model checkpoint**

### Chat inference

- [Chat inference](https://github.com/konpep-dev/aether-rwkv-/blob/main/inference.py): `inference.py`

### RWKV engine

- [RWKV model](https://github.com/konpep-dev/aether-rwkv-/blob/main/model.py): `model.py`
- [Time-mix recurrence](https://github.com/konpep-dev/aether-rwkv-/blob/main/model.py): `model.py`
- [Channel-mix layer](https://github.com/konpep-dev/aether-rwkv-/blob/main/model.py): `model.py`
- [WKV acceleration](https://github.com/konpep-dev/aether-rwkv-/blob/main/wkv_cuda_kernel.py): `wkv_cuda_kernel.py`

### Other components

- **Chat user**

## Connections

- Corpus generator → Synthetic corpus: generates (evidence: `README.md`)
- Synthetic corpus → Training loop: training data
- Training loop → Byte-BPE tokenizer: encodes samples (evidence: `train.py`)
- Training loop → RWKV model: trains (evidence: `train.py`)
- Training loop → Model checkpoint: saves weights (evidence: `train.py`)
- Model checkpoint → Chat inference: loads weights
- Chat user → Chat inference: submits prompts (evidence: `README.md`)
- Chat inference → Byte-BPE tokenizer: encodes and decodes (evidence: `inference.py`)
- Chat inference → RWKV model: generates tokens (evidence: `inference.py`)
- RWKV model → Time-mix recurrence: applies (evidence: `model.py`)
- RWKV model → Channel-mix layer: applies (evidence: `model.py`)
- Time-mix recurrence → WKV acceleration: dispatches WKV (evidence: `model.py`)
