# BNNSGraph.Context A wrapper around a compiled graph object that adds a required modifiable context to support dynamically sized models and set execute-time options. ``` class Context ``` ## Overview A [`BNNSGraph.Context`](/documentation/Accelerate/BNNSGraph/Context) instance provides a wrapper around the C API [`bnns_graph_t`](/documentation/Accelerate/bnns_graph_t) and [`bnns_graph_context_t`](/documentation/Accelerate/bnns_graph_context_t) types. Because this class manages its own memory, you don’t need to call [`BNNSGraphContextDestroy(_:)`](/documentation/Accelerate/BNNSGraphContextDestroy(_:)) to deallocate its resources. ## Topics ### Creating a graph context [`struct CompileOptions`](/documentation/Accelerate/BNNSGraph/CompileOptions) The compilation options that BNNS uses when compiling a source mlmodelc file to a graph object. ### Specifying and querying a graph context’s properties [`func setDynamicShapes([BNNSGraph.Shape], forFunction: String?) async throws -> [BNNSGraph.Shape]`](/documentation/Accelerate/BNNSGraph/Context/setDynamicShapes(_:forFunction:)) Specifies the dynamic shapes for a graph and, if possible, infers the output shapes. [`struct Shape`](/documentation/Accelerate/BNNSGraph/Shape) The specification of the shape of an argument. [`func argumentCount(forFunction: String?) -> Int`](/documentation/Accelerate/BNNSGraph/Context/argumentCount(forFunction:)) Returns the number of arguments for the given function argument. [`func argumentNames(forFunction: String?) -> [String]`](/documentation/Accelerate/BNNSGraph/Context/argumentNames(forFunction:)) Returns the names of arguments for the given function argument. [`func argumentPosition(forFunction: String?, argument: String) -> Int`](/documentation/Accelerate/BNNSGraph/Context/argumentPosition(forFunction:argument:)) Returns the index into the arguments array for the given function argument. [`var functionCount: Int`](/documentation/Accelerate/BNNSGraph/Context/functionCount) The number of input arguments for the given function argument. [`var functionNames: [String]`](/documentation/Accelerate/BNNSGraph/Context/functionNames) Returns the names of callable functions in the graph. [`var checkForNaNsAndInfinities: Bool`](/documentation/Accelerate/BNNSGraph/Context/checkForNaNsAndInfinities) A Boolean value that specifies that the context checks intermediate tensors for NaNs and infinities. ### Specifying a tensor’s properties [`func tensor(forFunction: String?, argument: String, fillKnownDynamicShapes: Bool) -> BNNSTensor?`](/documentation/Accelerate/BNNSGraph/Context/tensor(forFunction:argument:fillKnownDynamicShapes:)) Returns an unallocated tensor for a given function argument. ### Executing a graph [`func executeFunction(String?, arguments: inout [BNNSTensor]) async throws`](/documentation/Accelerate/BNNSGraph/Context/executeFunction(_:arguments:)-8bhcn) Executes the specified function using an array of input and output tensors. [`func executeFunction(String?, arguments: [T]) throws`](/documentation/Accelerate/BNNSGraph/Context/executeFunction(_:arguments:)-95snr) Executes the specified function using an array of input and output pointers. [`protocol PointerArgument`](/documentation/Accelerate/BNNSGraph/PointerArgument) A type that BNNS Graph accepts as an input-output argument. ### Handling errors [`enum Error`](/documentation/Accelerate/BNNSGraph/Error) Error codes that a graph context throws. ### Initializers [`init(compileFromPath: String, functionName: String?, options: BNNSGraph.CompileOptions) async throws`](/documentation/Accelerate/BNNSGraph/Context/init(compileFromPath:functionName:options:)-3nn5g) Returns a new context that wraps a graph object which represents the compiled `.mlmodelc` file. [`init(compileFromPath: String, functionName: String?, options: BNNSGraph.CompileOptions) throws`](/documentation/Accelerate/BNNSGraph/Context/init(compileFromPath:functionName:options:)-6ghot) Synchronously returns a new context that wraps a graph object which represents the compiled `.mlmodelc` file. ### Instance Properties [`var streamingAdvanceCount: Int`](/documentation/Accelerate/BNNSGraph/Context/streamingAdvanceCount) Sets streaming advancement amount for cases with dynamically shaped inputs. ### Instance Methods [`func allocateTensor(forFunction: String?, argument: String, fillKnownDynamicShapes: Bool) -> BNNSTensor?`](/documentation/Accelerate/BNNSGraph/Context/allocateTensor(forFunction:argument:fillKnownDynamicShapes:)) Returns an allocated tensor for a given function argument. [`func executeFunction(String?, arguments: inout [BNNSTensor]) throws`](/documentation/Accelerate/BNNSGraph/Context/executeFunction(_:arguments:)-52eh3) Synchronously executes the specified function with the provided context. [`func setBatchSize(Int, forFunction: String?) async`](/documentation/Accelerate/BNNSGraph/Context/setBatchSize(_:forFunction:)-4872x) Sets the batch size for a graph. [`func setBatchSize(Int, forFunction: String?)`](/documentation/Accelerate/BNNSGraph/Context/setBatchSize(_:forFunction:)-8eqzm) Synchronously sets the batch size for a graph. --- Copyright © 2026 Apple Inc. All rights reserved. | [Terms of Use](https://www.apple.com/legal/internet-services/terms/site.html) | [Privacy Policy](https://www.apple.com/privacy/privacy-policy)