# dexmal/dexbotic architecture

> Architecture diagram of the GitHub repository dexmal/dexbotic, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-09-25). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/dexmal/dexbotic
- Repository: https://github.com/dexmal/dexbotic

## Overview

Dexbotic is a PyTorch toolbox for developing vision-language-action systems: users prepare robot or simulation data, train or fine-tune a selected policy, then run inference and evaluate or deploy it on a robot. The tree documents a broad family of manipulation and navigation models, plus supervised, co-training, and RL paths. The excerpts directly show multimodal model construction, tokenization, an inference client, and RL rollout/training behavior; connections for other listed subsystems reflect README-described capabilities and the repository layout, not inspected implementations.

Main entry points: [`dexbotic/exp`](https://github.com/dexmal/dexbotic/tree/main/dexbotic/exp) (Experiment configs), [`dexbotic/exp/rl/rl_trainer.py`](https://github.com/dexmal/dexbotic/blob/main/dexbotic/exp/rl/rl_trainer.py) (RL trainer), [`dexbotic/model`](https://github.com/dexmal/dexbotic/tree/main/dexbotic/model) (VLA model families), [`dexbotic/client.py`](https://github.com/dexmal/dexbotic/blob/main/dexbotic/client.py) (Robot inference client)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_data["Data preparation"]
  node_sources["Dataset adapters"]
  node_dataset["Dataset pipeline"]
  node_collator["Batch collation<br/>[collator.py]"]
  node_tokenization["Conversation tokens<br/>[process.py]"]
end

subgraph group_training["Training"]
  node_experiments["Experiment configs"]
  node_trainer["Training engine<br/>[trainer.py]"]
end

subgraph group_models["VLA models"]
  node_modelcore["Multimodal core<br/>[dexbotic_arch.py]"]
  node_modelzoo["VLA model families"]
end

subgraph group_deployment["Inference deployment"]
  node_policy["Policy wrappers"]
  node_inference["Inference runtime<br/>[runtime.py]"]
  node_api["Robot inference client<br/>[client.py]"]
  node_hardware["Robot bridges"]
  node_simulation["Simulation environments"]
  node_evaluation["Policy evaluation"]
end

subgraph group_rl["RL post-training"]
  node_rltrainer["RL trainer<br/>[rl_trainer.py]"]
  node_rollouts["RL rollouts"]
  node_rlbridge["RL backend bridge"]
end

node_researcher(("Researcher"))
node_robot(("Robot operator"))
node_robotdata[("Robot datasets")]

node_researcher -->|"configures runs"| node_experiments
node_robotdata -->|"loads from"| node_sources
node_sources -->|"provides records"| node_dataset
node_dataset -->|"prepares samples"| node_collator
node_collator -->|"batches text"| node_tokenization
node_experiments -->|"configures training"| node_trainer
node_trainer -->|"trains"| node_modelzoo
node_modelzoo -->|"uses multimodal core"| node_modelcore
node_experiments -->|"selects data"| node_dataset
node_robot -->|"sends observations"| node_api
node_api -->|"requests predictions"| node_inference
node_inference -->|"dispatches inference"| node_policy
node_policy -->|"runs model"| node_modelzoo
node_api -->|"returns actions"| node_robot
node_hardware -->|"bridges robot I/O"| node_inference
node_simulation -->|"provides episodes"| node_evaluation
node_evaluation -->|"evaluates"| node_policy
node_experiments -->|"configures RL"| node_rltrainer
node_rltrainer -->|"collects trajectories"| node_rollouts
node_rollouts -->|"steps environments"| node_simulation
node_rltrainer -->|"updates policy"| node_modelzoo
node_rlbridge -.->|"integrates backend"| node_rltrainer

click node_sources "https://github.com/dexmal/dexbotic/tree/main/dexbotic/data/data_source"
click node_dataset "https://github.com/dexmal/dexbotic/tree/main/dexbotic/data/dataset"
click node_collator "https://github.com/dexmal/dexbotic/blob/main/dexbotic/data/collator.py"
click node_tokenization "https://github.com/dexmal/dexbotic/blob/main/dexbotic/tokenization/process.py"
click node_experiments "https://github.com/dexmal/dexbotic/tree/main/dexbotic/exp"
click node_trainer "https://github.com/dexmal/dexbotic/blob/main/dexbotic/exp/trainer.py"
click node_modelcore "https://github.com/dexmal/dexbotic/blob/main/dexbotic/model/dexbotic_arch.py"
click node_modelzoo "https://github.com/dexmal/dexbotic/tree/main/dexbotic/model"
click node_policy "https://github.com/dexmal/dexbotic/tree/main/dexbotic/policy"
click node_inference "https://github.com/dexmal/dexbotic/blob/main/dexbotic/infer/runtime.py"
click node_api "https://github.com/dexmal/dexbotic/blob/main/dexbotic/client.py"
click node_hardware "https://github.com/dexmal/dexbotic/tree/main/hardware"
click node_simulation "https://github.com/dexmal/dexbotic/tree/main/dexbotic/sim_envs"
click node_evaluation "https://github.com/dexmal/dexbotic/tree/main/dexbotic/exp"
click node_rltrainer "https://github.com/dexmal/dexbotic/blob/main/dexbotic/exp/rl/rl_trainer.py"
click node_rollouts "https://github.com/dexmal/dexbotic/blob/main/dexbotic/exp/rl/rl_rollout_redis.py"
click node_rlbridge "https://github.com/dexmal/dexbotic/tree/main/dexbotic/rl/rlinf_bridge"

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_sources,node_dataset,node_collator,node_tokenization toneBlue
class node_experiments,node_trainer,node_robotdata toneAmber
class node_modelcore,node_modelzoo toneMint
class node_policy,node_inference,node_api,node_hardware,node_simulation,node_evaluation toneRose
class node_rltrainer,node_rollouts,node_rlbridge,node_researcher,node_robot toneIndigo
```

## Components

### Data preparation

- [Dataset adapters](https://github.com/dexmal/dexbotic/tree/main/dexbotic/data/data_source): `dexbotic/data/data_source`
- [Dataset pipeline](https://github.com/dexmal/dexbotic/tree/main/dexbotic/data/dataset): `dexbotic/data/dataset`
- [Batch collation](https://github.com/dexmal/dexbotic/blob/main/dexbotic/data/collator.py): `dexbotic/data/collator.py`
- [Conversation tokens](https://github.com/dexmal/dexbotic/blob/main/dexbotic/tokenization/process.py): `dexbotic/tokenization/process.py`

### Training

- [Experiment configs](https://github.com/dexmal/dexbotic/tree/main/dexbotic/exp): `dexbotic/exp`
- [Training engine](https://github.com/dexmal/dexbotic/blob/main/dexbotic/exp/trainer.py): `dexbotic/exp/trainer.py`

### VLA models

- [Multimodal core](https://github.com/dexmal/dexbotic/blob/main/dexbotic/model/dexbotic_arch.py): `dexbotic/model/dexbotic_arch.py`
- [VLA model families](https://github.com/dexmal/dexbotic/tree/main/dexbotic/model): `dexbotic/model`

### Inference deployment

- [Policy wrappers](https://github.com/dexmal/dexbotic/tree/main/dexbotic/policy): `dexbotic/policy`
- [Inference runtime](https://github.com/dexmal/dexbotic/blob/main/dexbotic/infer/runtime.py): `dexbotic/infer/runtime.py`
- [Robot inference client](https://github.com/dexmal/dexbotic/blob/main/dexbotic/client.py): `dexbotic/client.py`
- [Robot bridges](https://github.com/dexmal/dexbotic/tree/main/hardware): `hardware`
- [Simulation environments](https://github.com/dexmal/dexbotic/tree/main/dexbotic/sim_envs): `dexbotic/sim_envs`
- [Policy evaluation](https://github.com/dexmal/dexbotic/tree/main/dexbotic/exp): `dexbotic/exp`

### RL post-training

- [RL trainer](https://github.com/dexmal/dexbotic/blob/main/dexbotic/exp/rl/rl_trainer.py): `dexbotic/exp/rl/rl_trainer.py`
- [RL rollouts](https://github.com/dexmal/dexbotic/blob/main/dexbotic/exp/rl/rl_rollout_redis.py): `dexbotic/exp/rl/rl_rollout_redis.py`
- [RL backend bridge](https://github.com/dexmal/dexbotic/tree/main/dexbotic/rl/rlinf_bridge): `dexbotic/rl/rlinf_bridge`

### Other components

- **Researcher**
- **Robot operator**
- **Robot datasets**

## Connections

- Researcher → Experiment configs: configures runs
- Robot datasets → Dataset adapters: loads from
- Dataset adapters → Dataset pipeline: provides records
- Dataset pipeline → Batch collation: prepares samples
- Batch collation → Conversation tokens: batches text
- Experiment configs → Training engine: configures training
- Training engine → VLA model families: trains
- VLA model families → Multimodal core: uses multimodal core
- Experiment configs → Dataset pipeline: selects data
- Robot operator → Robot inference client: sends observations
- Robot inference client → Inference runtime: requests predictions
- Inference runtime → Policy wrappers: dispatches inference
- Policy wrappers → VLA model families: runs model
- Robot inference client → Robot operator: returns actions
- Robot bridges → Inference runtime: bridges robot I/O
- Simulation environments → Policy evaluation: provides episodes
- Policy evaluation → Policy wrappers: evaluates
- Experiment configs → RL trainer: configures RL
- RL trainer → RL rollouts: collects trajectories
- RL rollouts → Simulation environments: steps environments
- RL trainer → VLA model families: updates policy
- RL backend bridge → RL trainer: integrates backend