# nvidia/deeplearningexamples architecture

> Architecture diagram of the GitHub repository nvidia/deeplearningexamples, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-10). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/nvidia/deeplearningexamples
- Repository: https://github.com/nvidia/deeplearningexamples

## Overview

This is NVIDIA’s collection of deep-learning examples, not one application. Its visible runtime paths cover model training and inference across multiple frameworks; the sampled FastSpeech code gives the clearest internal flow, including TensorRT inference. The README excerpt documents broad computer-vision capabilities, but does not expose their implementation or wiring. The graph therefore preserves prominent indexed modules and documented capabilities while omitting relationships that the available excerpts do not establish.

Main entry points: [`CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_inferencer.py`](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_inferencer.py) (TensorRT inference), [`CUDA-Optimized/FastSpeech/fastspeech/trt/trt_inferencer.py`](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/trt_inferencer.py) (TensorRT base), [`CUDA-Optimized/FastSpeech/fastspeech/utils/logging.py`](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/utils/logging.py) (Timestamp logging), [`CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_multi_engine_inferencer.py`](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_multi_engine_inferencer.py) (Multi-engine inference)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_speech["Speech synthesis"]
  node_fastspeech_train["FastSpeech training<br/>[train.py]"]
  node_fastspeech_infer["FastSpeech inference<br/>[infer.py]"]
  node_fastspeech_trt["TensorRT inference"]
  node_multi_engine["Multi-engine inference"]
  node_trt_common["TensorRT buffers<br/>[common.py]"]
  node_trt_base["TensorRT base<br/>[trt_inferencer.py]"]
  node_speech_output["Generated speech<br/>[audio.py]"]
end

subgraph group_vision["Vision examples"]
  node_se3_data["QM9 data loading<br/>[qm9.py]"]
  node_se3_model["SE3 transformer<br/>[transformer.py]"]
  node_mxnet_models["MXNet classifiers<br/>[models.py]"]
  node_paddle_resnet["Paddle ResNet<br/>[resnet.py]"]
  node_paddle_program["Paddle training program<br/>[program.py]"]
end

subgraph group_language["Language models"]
  node_paddle_bert_program["Paddle BERT program<br/>[program.py]"]
  node_paddle_bert_loss["BERT loss<br/>[loss.py]"]
  node_paddle_bert_scheduler["BERT scheduler<br/>[lr_scheduler.py]"]
  node_paddle_bert_optimizer["BERT optimizer<br/>[optimizer.py]"]
end

subgraph group_deployment["Triton deployment"]
  node_gpunet_pipeline["GPUNet deployment stages<br/>[pipeline.py]"]
  node_tft_pipeline["TFT deployment stages<br/>[pipeline.py]"]
  node_bert_pipeline["BERT deployment stages<br/>[pipeline.py]"]
end

subgraph group_shared["Shared runtime"]
  node_fastspeech_logging["Timestamp logging<br/>[logging.py]"]
  node_fastspeech_time["Elapsed-time helper<br/>[time.py]"]
  node_fastspeech_nvtx["NVTX instrumentation<br/>[nvtx.py]"]
  node_fastspeech_pytorch["PyTorch utilities<br/>[pytorch.py]"]
end

node_speech_user(("Speech user"))

node_speech_user -.->|"requests synthesis"| node_fastspeech_infer
node_fastspeech_trt -->|"creates buffers"| node_trt_common
node_fastspeech_trt -->|"uses base"| node_trt_base
node_multi_engine -->|"extends inferencer"| node_fastspeech_trt
node_trt_base -->|"logs messages"| node_fastspeech_logging
node_fastspeech_time -->|"prints timing"| node_fastspeech_logging
node_gpunet_pipeline -.->|"similar stages"| node_tft_pipeline

click node_fastspeech_train "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/train.py"
click node_fastspeech_infer "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/infer.py"
click node_fastspeech_trt "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_inferencer.py"
click node_multi_engine "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_multi_engine_inferencer.py"
click node_trt_common "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/common.py"
click node_trt_base "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/trt_inferencer.py"
click node_speech_output "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/audio.py"
click node_se3_data "https://github.com/nvidia/deeplearningexamples/blob/master/DGLPyTorch/DrugDiscovery/SE3Transformer/se3_transformer/data_loading/qm9.py"
click node_se3_model "https://github.com/nvidia/deeplearningexamples/blob/master/DGLPyTorch/DrugDiscovery/SE3Transformer/se3_transformer/model/transformer.py"
click node_mxnet_models "https://github.com/nvidia/deeplearningexamples/blob/master/MxNet/Classification/RN50v1.5/models.py"
click node_paddle_resnet "https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/Classification/RN50v1.5/models/resnet.py"
click node_paddle_program "https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/Classification/RN50v1.5/program.py"
click node_paddle_bert_program "https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/LanguageModeling/BERT/program.py"
click node_paddle_bert_loss "https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/LanguageModeling/BERT/loss.py"
click node_paddle_bert_scheduler "https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/LanguageModeling/BERT/lr_scheduler.py"
click node_paddle_bert_optimizer "https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/LanguageModeling/BERT/optimizer.py"
click node_gpunet_pipeline "https://github.com/nvidia/deeplearningexamples/blob/master/PyTorch/Classification/GPUNet/triton/runner/pipeline.py"
click node_tft_pipeline "https://github.com/nvidia/deeplearningexamples/blob/master/PyTorch/Forecasting/TFT/triton/runner/pipeline.py"
click node_bert_pipeline "https://github.com/nvidia/deeplearningexamples/blob/master/PyTorch/LanguageModeling/BERT/triton/runner/pipeline.py"
click node_fastspeech_logging "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/utils/logging.py"
click node_fastspeech_time "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/utils/time.py"
click node_fastspeech_nvtx "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/utils/nvtx.py"
click node_fastspeech_pytorch "https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/utils/pytorch.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_fastspeech_train,node_fastspeech_infer,node_fastspeech_trt,node_multi_engine,node_trt_common,node_trt_base,node_speech_output,node_speech_user toneBlue
class node_se3_data,node_se3_model,node_mxnet_models,node_paddle_resnet,node_paddle_program toneAmber
class node_paddle_bert_program,node_paddle_bert_loss,node_paddle_bert_scheduler,node_paddle_bert_optimizer toneMint
class node_gpunet_pipeline,node_tft_pipeline,node_bert_pipeline toneRose
class node_fastspeech_logging,node_fastspeech_time,node_fastspeech_nvtx,node_fastspeech_pytorch toneIndigo
```

## Components

### Speech synthesis

- [FastSpeech training](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/train.py): `CUDA-Optimized/FastSpeech/fastspeech/train.py`
- [FastSpeech inference](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/infer.py): `CUDA-Optimized/FastSpeech/fastspeech/infer.py`
- [TensorRT inference](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_inferencer.py): `CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_inferencer.py`
- [Multi-engine inference](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_multi_engine_inferencer.py): `CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_multi_engine_inferencer.py`
- [TensorRT buffers](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/common.py): `CUDA-Optimized/FastSpeech/fastspeech/trt/common.py`
- [TensorRT base](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/trt/trt_inferencer.py): `CUDA-Optimized/FastSpeech/fastspeech/trt/trt_inferencer.py`
- [Generated speech](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/audio.py): `CUDA-Optimized/FastSpeech/fastspeech/audio.py`

### Vision examples

- [QM9 data loading](https://github.com/nvidia/deeplearningexamples/blob/master/DGLPyTorch/DrugDiscovery/SE3Transformer/se3_transformer/data_loading/qm9.py): `DGLPyTorch/DrugDiscovery/SE3Transformer/se3_transformer/data_loading/qm9.py`
- [SE3 transformer](https://github.com/nvidia/deeplearningexamples/blob/master/DGLPyTorch/DrugDiscovery/SE3Transformer/se3_transformer/model/transformer.py): `DGLPyTorch/DrugDiscovery/SE3Transformer/se3_transformer/model/transformer.py`
- [MXNet classifiers](https://github.com/nvidia/deeplearningexamples/blob/master/MxNet/Classification/RN50v1.5/models.py): `MxNet/Classification/RN50v1.5/models.py`
- [Paddle ResNet](https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/Classification/RN50v1.5/models/resnet.py): `PaddlePaddle/Classification/RN50v1.5/models/resnet.py`
- [Paddle training program](https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/Classification/RN50v1.5/program.py): `PaddlePaddle/Classification/RN50v1.5/program.py`

### Language models

- [Paddle BERT program](https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/LanguageModeling/BERT/program.py): `PaddlePaddle/LanguageModeling/BERT/program.py`
- [BERT loss](https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/LanguageModeling/BERT/loss.py): `PaddlePaddle/LanguageModeling/BERT/loss.py`
- [BERT scheduler](https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/LanguageModeling/BERT/lr_scheduler.py): `PaddlePaddle/LanguageModeling/BERT/lr_scheduler.py`
- [BERT optimizer](https://github.com/nvidia/deeplearningexamples/blob/master/PaddlePaddle/LanguageModeling/BERT/optimizer.py): `PaddlePaddle/LanguageModeling/BERT/optimizer.py`

### Triton deployment

- [GPUNet deployment stages](https://github.com/nvidia/deeplearningexamples/blob/master/PyTorch/Classification/GPUNet/triton/runner/pipeline.py): `PyTorch/Classification/GPUNet/triton/runner/pipeline.py`
- [TFT deployment stages](https://github.com/nvidia/deeplearningexamples/blob/master/PyTorch/Forecasting/TFT/triton/runner/pipeline.py): `PyTorch/Forecasting/TFT/triton/runner/pipeline.py`
- [BERT deployment stages](https://github.com/nvidia/deeplearningexamples/blob/master/PyTorch/LanguageModeling/BERT/triton/runner/pipeline.py): `PyTorch/LanguageModeling/BERT/triton/runner/pipeline.py`

### Shared runtime

- [Timestamp logging](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/utils/logging.py): `CUDA-Optimized/FastSpeech/fastspeech/utils/logging.py`
- [Elapsed-time helper](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/utils/time.py): `CUDA-Optimized/FastSpeech/fastspeech/utils/time.py`
- [NVTX instrumentation](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/utils/nvtx.py): `CUDA-Optimized/FastSpeech/fastspeech/utils/nvtx.py`
- [PyTorch utilities](https://github.com/nvidia/deeplearningexamples/blob/master/CUDA-Optimized/FastSpeech/fastspeech/utils/pytorch.py): `CUDA-Optimized/FastSpeech/fastspeech/utils/pytorch.py`

### Other components

- **Speech user**

## Connections

- Speech user → FastSpeech inference: requests synthesis
- TensorRT inference → TensorRT buffers: creates buffers (evidence: `CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_inferencer.py`)
- TensorRT inference → TensorRT base: uses base (evidence: `CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_inferencer.py`)
- Multi-engine inference → TensorRT inference: extends inferencer (evidence: `CUDA-Optimized/FastSpeech/fastspeech/trt/fastspeech_trt_multi_engine_inferencer.py`)
- TensorRT base → Timestamp logging: logs messages (evidence: `CUDA-Optimized/FastSpeech/fastspeech/trt/trt_inferencer.py`)
- Elapsed-time helper → Timestamp logging: prints timing (evidence: `CUDA-Optimized/FastSpeech/fastspeech/utils/time.py`)
- GPUNet deployment stages → TFT deployment stages: similar stages
