Model Configuration

Warning

You are currently viewing an out-of-date version of the Triton documentation. For the latest documentation visit the Triton documentation on GitHub.

Each model in a Model Repository must include a model configuration that provides required and optional information about the model. Typically, this configuration is provided in a config.pbtxt file specified as ModelConfig protobuf. In some cases, discussed in Generated Model Configuration, the model configuration can be generated automatically by Triton and so does not need to be provided explicitly.

Minimal Model Configuration

A minimal model configuration must specify platform, max_batch_size, input, and output.

As an example consider a TensorRT model that has two inputs, input0 and input1, and one output, output0, all of which are 16 entry float32 tensors. The minimal configuration is:

platform: "tensorrt_plan"
max_batch_size: 8
input [
  {
    name: "input0"
    data_type: TYPE_FP32
    dims: [ 16 ]
  },
  {
    name: "input1"
    data_type: TYPE_FP32
    dims: [ 16 ]
  }
]
output [
  {
    name: "output0"
    data_type: TYPE_FP32
    dims: [ 16 ]
  }
]

Name and Platform

If the name of the model is not specified in the configuration it is assumed to be the basename of the model repository directory containing the model. If name is specified it must match the basename of the model repository directory containing the model. The platform must be one of tensorrt_plan, tensorflow_graphdef, tensorflow_savedmodel, caffe2_netdef, onnxruntime_onnx, pytorch_libtorch or custom.

Maximum Batch Size

The max_batch_size value indicates the maximum batch size that the model supports for the type of batching that can be exploited by Triton. If the model’s batch dimension is the first dimension, and all inputs and outputs to the model have this batch dimension, then Triton can use its Dynamic Batcher or Sequence Batcher to automatically use batching with the model. In this case max_batch_size should be set to a value >=1 that indicates the maximum batch size that Triton should use with the model.

For models that do not support batching, or do not support batching in the specific was described above, max_batch_size must be set to zero.

Inputs and Outputs

Each model input and output must specify a name, datatype, and shape.

The name specified for an input or output tensor must match the name expected by the model. PyTorch Naming Convention: Due to the absence of names for inputs and outputs in a TorchScript model, the “name” attribute of both the inputs and outputs in the configuration must follow a specific naming convention i.e. “<name>__<index>”. Where <name> can be any string and <index> refers to the position of the corresponding input/output. This means if there are two inputs and two outputs they must be named as: “INPUT__0”, “INPUT__1” and “OUTPUT__0”, “OUTPUT__1” such that “INPUT__0” refers to first input and INPUT__1 refers to the second input, etc.

The datatypes allowed for input and output tensors varies based on the type of the model. Section Datatypes describes the allowed datatypes and how they map to the datatypes of each model type.

An input shape indicates the shape of an input tensor expected by the model and by Triton in inference requests. An output shape indicates the shape of an output tensor produced by the model and returned by Triton in response to an inference request. Both input and output shape must have rank >= 1, that is, the empty shape [ ] is not allowed.

Input and output shapes are specified by a combination of max_batch_size and the dimensions specified by input dims or output dims. For models with max_batch_size > 0, the full shape is formed as [ -1 ] + <dims>, where <dims> is the shape specified by by input dims or output dims. For models with max_batch_size == 0, the full shape is formed as <dims>. For example, for the following configuration the shape of “input0” is [ -1, 16 ] and the shape of “output0” is [ -1, 4 ]:

platform: "tensorrt_plan"
max_batch_size: 8
input [
  {
    name: "input0"
    data_type: TYPE_FP32
    dims: [ 16 ]
  }
]
output [
  {
    name: "output0"
    data_type: TYPE_FP32
    dims: [ 4 ]
  }
]

For a configuration that is identical except that max_batch_size == 0, the shape of “input0” is [ 16 ] and the shape of “output0” is [ 4 ]:

platform: "tensorrt_plan"
max_batch_size: 0
input [
  {
    name: "input0"
    data_type: TYPE_FP32
    dims: [ 16 ]
  }
]
output [
  {
    name: "output0"
    data_type: TYPE_FP32
    dims: [ 4 ]
  }
]

For models that support input and output tensors with variable-size dimensions, those dimensions can be listed as -1 in the input and output configuration. For example, if a model requires a 2-dimensional input tensor where the first dimension must be size 4 but the second dimension can be any size, the model configuration for that input would include dims: [ 4, -1 ]. Triton would then accept inference requests where that input tensor’s second dimension was any value >= 0. The model configuration can be more restrictive than what is allowed by the underlying model. For example, even though the model allows the second dimension to be any size, the model configuration could be specific as dims: [ 4, 4 ]. In this case, Triton would only accept inference requests where the input tensor’s shape was exactly [ 4, 4 ].

The reshape property must be used if there is a mismatch between the input shape that Triton receives in an inference request and the input shape expected by the model. Similarly, the reshape property must be used if there is a mismatch between the output shape produced by the model and the shape that Triton returns in a response to an inference request.

Generated Model Configuration

By default, the model configuration file containing the required settings must be provided with each model. However, if Triton is started with the --strict-model-config=false option, then in some cases the required portions of the model configuration file can be generated automatically by Triton. The required portion of the model configuration are those settings shown in the example minimal configuration above. Specifically:

  • TensorRT Plan models do not require a model configuration file because Triton can derive all the required settings automatically.

  • TensorFlow SavedModel models do not require a model configuration file because Triton can derive all the required settings automatically.

  • ONNX Runtime ONNX models do not require a model configuration file because Triton can derive all the required settings automatically.

When using --strict-model-config=false you can see the model configuration that was generated for a model by using the metadata endpoint.

Triton only generates the required portion of the model configuration file. You must still provide the optional portions of the model configuration if necessary, such as version_policy, optimization, scheduling and batching, instance_group, default_model_filename, cc_model_filenames, and tags.

When serving a classification model, keep in mind that label_filename cannot be automatically derived. You will need to either create a config.pbtxt file specifying all required output along with the label_filename, or handle the mapping from model output to label in the client code directly.

Datatypes

The following table shows the tensor datatypes supported by Triton. The first column shows the name of the datatype as it appears in the model configuration file. The other columns show the corresponding datatype for the model frameworks and for the Python numpy library. If a model framework does not have an entry for a given datatype, then Triton does not support that datatype for that model.

Type

TensorRT

TensorFlow

Caffe2

ONNX Runtime

PyTorch

NumPy

TYPE_BOOL

kBOOL

DT_BOOL

BOOL

BOOL

kBool

bool

TYPE_UINT8

DT_UINT8

UINT8

UINT8

kByte

uint8

TYPE_UINT16

DT_UINT16

UINT16

UINT16

uint16

TYPE_UINT32

DT_UINT32

UINT32

uint32

TYPE_UINT64

DT_UINT64

UINT64

uint64

TYPE_INT8

kINT8

DT_INT8

INT8

INT8

kChar

int8

TYPE_INT16

DT_INT16

INT16

INT16

kShort

int16

TYPE_INT32

kINT32

DT_INT32

INT32

INT32

kInt

int32

TYPE_INT64

DT_INT64

INT64

INT64

kLong

int64

TYPE_FP16

kHALF

DT_HALF

FLOAT16

FLOAT16

float16

TYPE_FP32

kFLOAT

DT_FLOAT

FLOAT

FLOAT

kFloat

float32

TYPE_FP64

DT_DOUBLE

DOUBLE

DOUBLE

kDouble

float64

TYPE_STRING

DT_STRING

STRING

dtype(object)

For TensorRT each value is in the nvinfer1::DataType namespace. For example, nvinfer1::DataType::kFLOAT is the 32-bit floating-point datatype.

For TensorFlow each value is in the tensorflow namespace. For example, tensorflow::DT_FLOAT is the 32-bit floating-point value.

For Caffe2 each value is in the caffe2 namespace and is prepended with TensorProto_DataType_. For example, caffe2::TensorProto_DataType_FLOAT is the 32-bit floating-point datatype.

For ONNX Runtime each value is prepended with ONNX_TENSOR_ELEMENT_DATA_TYPE_. For example, ONNX_TENSOR_ELEMENT_DATA_TYPE_FLOAT is the 32-bit floating-point datatype.

For PyTorch each value is in the torch namespace. For example, torch::kFloat is the 32-bit floating-point datatype.

For Numpy each value is in the numpy module. For example, numpy.float32 is the 32-bit floating-point datatype.

Reshape

The ModelTensorReshape property on a model configuration input or output is used to indicate that the input or output shape accepted by the inference API differs from the input or output shape expected or produced by the underlying framework model or custom backend.

For an input, reshape can be used to reshape the input tensor to a different shape expected by the framework or backend. A common use-case is where a model that supports batching expects a batched input to have shape [ batch-size ], which means that the batch dimension fully describes the shape. For the inference API the equivalent shape [ batch-size, 1 ] must be specified since each input in the batch must specify a non-empty shape. For this case the input should be specified as:

input [
  {
    name: "in"
    dims: [ 1 ]
    reshape: { shape: [ ] }
  }
  ...

For an output, reshape can be used to reshape the output tensor produced by the framework or backend to a different shape that is returned by the inference API. A common use-case is where a model that supports batching expects a batched output to have shape [ batch-size ], which means that the batch dimension fully describes the shape. For the inference API the equivalent shape [ batch-size, 1 ] must be specified since each output in the batch must specify a non-empty shape. For this case the output should be specified as:

output [
  {
    name: "in"
    dims: [ 1 ]
    reshape: { shape: [ ] }
  }
  ...

Shape Tensors

For models that support shape tensors, is_shape_tensor must be appropriately set for inputs and is_shape_tensor must be correctly set for outputs. Consider the following example configuration to understand how to use shape tensors with batching:

name: "myshapetensormodel"
platform: "tensorrt_plan"
max_batch_size: 8
input [
  {
    name: "input0"
    data_type: TYPE_FP32
    dims: [ -1 ]
  },
  {
    name: "input1"
    data_type: TYPE_INT32
    dims: [ 1 ]
    is_shape_tensor: true
  }
]
output [
  {
    name: "output0"
    data_type: TYPE_FP32
    dims: [ -1 ]
  }
]

As discussed before, Triton assumes that batching occurs along the first dimension which is not listed in in the input or output tensor dims. However, for shape tensors, batching occurs at the first shape value. For the above example, an inference request must provide inputs with the following shapes:

"input0": [ x, -1]
"input1": [ 1 ]
"output0": [ x, -1]

Where x is the batch size of the request. Triton requires the shape tensors to be marked as shape tensors in the model when using batching. Note that “input1” has shape [ 1 ] and not [ 2 ]. Triton will prepend the shape value x at “input1” before issuing the request to model.

Version Policy

Each model can have one or more versions available in the model repository. The nvidia::inferenceserver::ModelVersionPolicy schema allows the following policies.

  • All: All versions of the model that are available in the model repository are available for inferencing.

  • Latest: Only the latest ‘n’ versions of the model in the repository are available for inferencing. The latest versions of the model are the numerically greatest version numbers.

  • Specific: Only the specifically listed versions of the model are available for inferencing.

If no version policy is specified, then Latest (with num_version = 1) is used as the default, indicating that only the most recent version of the model is made available by Triton. In all cases, the addition or removal of version subdirectories from the model repository can change which model version is used on subsequent inference requests.

The following configuration specifies that all versions of the model will be available from the server:

platform: "tensorrt_plan"
max_batch_size: 8
input [
  {
    name: "input0"
    data_type: TYPE_FP32
    dims: [ 16 ]
  },
  {
    name: "input1"
    data_type: TYPE_FP32
    dims: [ 16 ]
  }
]
output [
  {
    name: "output0"
    data_type: TYPE_FP32
    dims: [ 16 ]
  }
]
version_policy: { all { }}

Instance Groups

Triton can provide multiple execution instances of a model so that multiple inference requests for that model can be handled simultaneously. The model configuration ModelInstanceGroup is used to specify the number of execution instances that should be made available and what compute resource should be used for those instances.

By default, a single execution instance of the model is created for each GPU available in the system. The instance-group setting can be used to place multiple execution instances of a model on every GPU or on only certain GPUs. For example, the following configuration will place two execution instances of the model to be available on each system GPU:

instance_group [
  {
    count: 2
    kind: KIND_GPU
  }
]

And the following configuration will place one execution instance on GPU 0 and two execution instances on GPUs 1 and 2:

instance_group [
  {
    count: 1
    kind: KIND_GPU
    gpus: [ 0 ]
  },
  {
    count: 2
    kind: KIND_GPU
    gpus: [ 1, 2 ]
  }
]

The instance group setting is also used to enable exection of a model on the CPU. A model can be executed on the CPU even if there is a GPU available in the system. The following places two execution instances on the CPU:

instance_group [
  {
    count: 2
    kind: KIND_CPU
  }
]

Scheduling And Batching

Triton supports batch inferencing by allowing individual inference requests to specify a batch of inputs. The inferencing for a batch of inputs is performed at the same time which is especially important for GPUs since it can greatly increase inferencing throughput. In many use cases the individual inference requests are not batched, therefore, they do not benefit from the throughput benefits of batching.

The inference server contains multiple scheduling and batching algorithms that support many different model types and use-cases. More information about model types and schedulers can be found in Models And Schedulers.

Default Scheduler

The default scheduler is used for a model if none of the scheduling_choice configurations are specified. This scheduler distributes inference requests to all instances configured for the model.

Dynamic Batcher

Dynamic batching is a feature of Triton that allows inference requests to be combined by the server, so that a batch is created dynamically, resulting in increased throughput. The dynamic batcher should be used for stateless models. The dynamically created batches are distributed to all instances configured for the model.

Dynamic batching is enabled and configured independently for each model using the ModelDynamicBatching settings in the model configuration. These settings control the preferred size(s) of the dynamically created batches, the maximum time that requests can be delayed in the scheduler to allow other requests to join the dynamic batch, and queue properties such a queue size, priorities, and time-outs.

Preferred Batch Sizes

The preferred_batch_size setting indicates the batch sizes that the dynamic batcher should attempt to create. For example, the following configuration enables dynamic batching with preferred batch sizes of 4 and 8:

dynamic_batching {
  preferred_batch_size: [ 4, 8 ]
}

When a model instance becomes available for inferencing, the dynamic batcher will attempt to create batches from the requests that are available in the scheduler. Requests are added to the batch in the order the requests were received. If the dynamic batcher can form a batch of a preferred size(s) it will create a batch of the largest possible preferred size and send it for inferencing. If the dynamic batcher cannot form a batch of a preferred size, it will send a batch of the largest size possible that is less than the max batch size allowed by the model. But see the following section for the delay option that changes this behavior.

The size of generated batches can be examined in aggregate using Count metrics, see Metrics. Triton verbose logging can be used to examine the size of individual batches.

Delayed Batching

The dynamic batcher can be configured to allow requests to be delayed for a limited time in the scheduler to allow other requests to join the dynamic batch. For example, the following configuration sets the maximum delay time of 100 microseconds for a request:

dynamic_batching {
  preferred_batch_size: [ 4, 8 ]
  max_queue_delay_microseconds: 100
}

The max_queue_delay_microseconds setting changes the dynamic batcher behavior when a batch of a preferred size cannot be created. When a batch of a preferred size cannot be created from the available requests, the dynamic batcher will delay sending the batch as long as no request is delayed longer than the configured max_queue_delay_microseconds setting. If a new request arrives during this delay and allows the dynamic batcher to form a batch of a preferred batch size, then that batch is sent immediately for inferencing. If the delay expires the dynamic batcher sends the batch as is, even though it is not a preferred size.

Preserve Ordering

The preserve_ordering setting is used to force all responses to be returned in the same order as requests were received. See the protobuf documentation for details.

Priority Levels

By default the dynamic batcher maintains a single queue that holds all inference requests for a model. The requests are processed and batched in order. The priority_levels setting can be used to create multiple priority levels within the dynamic batcher so that requests with higher priority are allowed to bypass requests with lower priority. Requests at the same priority level are processed in order. Inference requests that do not set a priority are scheduled using the default_priority_level.

Queue Policy

The dynamic batcher provides several settings that control how requests are queued for batching.

When priority_levels is not defined the ModelQueuePolicy for the single queue can be set with default_queue_policy.

When priority_levels is defined, each priority level can have a different ModelQueuePolicy as specified by default_queue_policy and priority_queue_policy.

The ModelQueuePolicy allows a maximum queue size to be set using the max_queue_size setting. The queue policy timeout_action, default_timeout_microseconds, and allow_timeout_override settings allow the queue to be configured so that individual requests are rejected or deferred if their time in the queue exceeds a specified timeout.

Sequence Batcher

Like the dynamic batcher, the sequence batcher combines non-batched inference requests, so that a batch is created dynamically. Unlike the dynamic batcher, the sequence batcher should be used for stateful models where a sequence of inference requests must be routed to the same model instance. The dynamically created batches are distributed to all instances configured for the model.

Sequence batching is enabled and configured independently for each model using the ModelSequenceBatching settings in the model configuration. These settings control the sequence timeout as well as configuring how Triton will send control signals to the model indicating sequence start, end, ready and correlation ID. See Models And Schedulers for more information and examples.

The size of generated batches can be examined in aggregate using Count metrics, see Metrics. Triton verbose logging can be used to examine the size of individual batches.

Ensemble Scheduler

The ensemble scheduler must be used for ensemble models and cannot be used for any other type of model.

The ensemble scheduler is enabled and configured independently for each model using the ModelEnsembleScheduling settings in the model configuration. The settings describe the models that are included in the ensemble and the flow of tensor values between the models. See Ensemble Models for more information and examples.

Optimization Policy

The model configuration ModelOptimizationPolicy is used to specify optimization and prioritization settings for a model. These settings control if/how a model is optimized by the backend framework and how it is scheduled and executed by Triton. See the protobuf documentation for the currently available settings.

TensorRT Optimization

TensorRT has existing integrations into the frameworks for TensorFlow and ONNX Runtime that can be easily enabled via Triton model configuration changes. When enabled, TensorRT optimizations will be applied to the TensorFlow or ONNX model at load time or when the model first receives inference requests. TensorRT optimizations include specializing and fusing model layers, and using reduced precision (for example, 16-bit floating-point) to provide significant throughput and latency improvements. These TensorRT optimizations occur inside the existing TensorFlow or ONNX models, transparent to the user.

Model Warmup

When a model is loaded by Triton the corresponding model framework initializes for that model. For some frameworks, some or all of this initialization is deferred until the model receives its first inference request (or first few inference requests). As a result, the first (few) inference requests can be significantly slower due to deferred initialization.

To avoid these initial, slow inference requests, Triton provides a configuration option that enables a model to be “warmed up” so that it is completely initialized before the first inference request is received. When the ModelWarmup option is used, Triton will not show the model as being ready for inference until model warmup has completed.

The model configuration ModelWarmup is used to specify warmup settings for a model. The settings define a series of inference requests that Triton will create to warm-up each model instance. A model instance will be served only if it completes the requests successfully. Note that the effect of warming up models varies depending on the framework backend, and it will cause Triton to be less responsive to model update, so the users should experiment and choose the configuration that suits their need. See the protobuf documentation for the currently available settings.