Struct GraphNode 

pub struct GraphNode {
    pub id: String,
    pub name: String,
    pub operation_type: OperationType,
    pub input_shapes: Vec<Vec<usize>>,
    pub output_shapes: Vec<Vec<usize>>,
    pub flop_count: Option<u64>,
    pub memory_usage: Option<u64>,
    pub execution_time_us: Option<u64>,
    pub parameter_count: Option<u64>,
    pub topo_order: Option<usize>,
    pub depth: usize,
    pub metadata: HashMap<String, String>,
}

Represents a single node in the computation graph

Fields

id: String

Unique identifier for this node

name: String

Human-readable name

operation_type: OperationType

Type of operation (e.g., “MatMul”, “Add”, “ReLU”)

input_shapes: Vec<Vec<usize>>

Input tensor shapes

output_shapes: Vec<Vec<usize>>

Output tensor shapes

flop_count: Option<u64>

Computational complexity (FLOPs), estimated from Self::input_shapes.

None when the node was built without shapes: FLOPs are a function of tensor extents, and nothing else here can supply them. Every node used to carry a constant here instead (1_000_000 for MatMul, 1_000 for elementwise ops, 5_000 for normalisations) because ComputationGraphAnalyzer::create_graph passed an empty shape slice to the estimator, making every shape-dependent branch unreachable.

memory_usage: Option<u64>

Memory usage estimate in bytes, from Self::input_shapes; None for the same reason as Self::flop_count (previously a constant 1024).

execution_time_us: Option<u64>

Execution time in microseconds (if profiled)

parameter_count: Option<u64>

Number of learned parameters for parameterized operations, derived from the node’s shapes; None for operations that have none, or when the shapes needed to count them are absent.

It used to return Some(1_000_000) for every MatMul, Some(500_000) for every Conv2D and Some(2_000_000) for every Embedding regardless of the model – literal “Example: 1M parameters” values.

topo_order: Option<usize>

Position in topological ordering

depth: usize

Depth in the graph (distance from inputs)

metadata: HashMap<String, String>

Additional metadata

Trait Implementations

impl Clone for GraphNode

fn clone(&self) -> GraphNode

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for GraphNode

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for GraphNode

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for GraphNode

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.