diff --git a/Conceptual_Guide/Part_5-Model_Ensembles/README.md b/Conceptual_Guide/Part_5-Model_Ensembles/README.md index 4bc24347..633173d8 100644 --- a/Conceptual_Guide/Part_5-Model_Ensembles/README.md +++ b/Conceptual_Guide/Part_5-Model_Ensembles/README.md @@ -189,7 +189,7 @@ ensemble_model/ ├── 1 └── config.pbtxt ``` -This time, we only need the configuration file to describe our ensemble along with an empty version folder (which you will need to create with `mkdir -p model_repository/ensemble_model/1`). Within the config file, we'll define the execution graph of our ensemble. This graph describes what the overall inputs and outputs of the ensemble will be, as well as how the data will flow through the models in the form of a Directed Acyclic Graph. Below is a graphical representation of our model pipeline. The diamonds represent the final input and output of the ensemble, which is all the client will interact with. The circles are the different deployed models, and the rectangles are the tensors that get passed between models. +This time, we only need the configuration file to describe our ensemble along with an empty version folder (which you can see at `model_repository/ensemble_model/1`). Within the config file, we'll define the execution graph of our ensemble. This graph describes what the overall inputs and outputs of the ensemble will be, as well as how the data will flow through the models in the form of a Directed Acyclic Graph. Below is a graphical representation of our model pipeline. The diamonds represent the final input and output of the ensemble, which is all the client will interact with. The circles are the different deployed models, and the rectangles are the tensors that get passed between models. ```mermaid flowchart LR @@ -362,4 +362,4 @@ python client.py You should see the parsed text printed out to your console. ## What's Next -In this example, we showed how you can use Model Ensembles to execute multiple models on Triton with a single network call. Model Ensembles are a great solution when your model pipelines are in the form of a Directed Acyclic Graph. However, not all pipelines can be expressed this way. For example, if your pipeline logic requires conditional branching or looped execution, you might need a more expressive way to define your pipeline. In the [next example](../Part_6-building_complex_pipelines/), we'll explore how you can create define more complex pipelines in Python using [Business Logic Scripting](https://github.com/triton-inference-server/python_backend#business-logic-scripting). \ No newline at end of file +In this example, we showed how you can use Model Ensembles to execute multiple models on Triton with a single network call. Model Ensembles are a great solution when your model pipelines are in the form of a Directed Acyclic Graph. However, not all pipelines can be expressed this way. For example, if your pipeline logic requires conditional branching or looped execution, you might need a more expressive way to define your pipeline. In the [next example](../Part_6-building_complex_pipelines/), we'll explore how you can create define more complex pipelines in Python using [Business Logic Scripting](https://github.com/triton-inference-server/python_backend#business-logic-scripting). diff --git a/Conceptual_Guide/Part_5-Model_Ensembles/model_repository/ensemble_model/1/.gitkeep b/Conceptual_Guide/Part_5-Model_Ensembles/model_repository/ensemble_model/1/.gitkeep new file mode 100644 index 00000000..e69de29b