Module model 

Model operations for saving, loading and inference

This module provides high-level operations for:

• Saving models in ONNX format
• Loading models from ONNX format
• Running inference on loaded models
• Training support (forward/backward passes)

Re-exports

pub use crate::llo::OnnxAttribute;

pub use crate::llo::OnnxModel;

pub use crate::llo::OnnxNode;

pub use crate::llo::OnnxTensor;

pub use crate::llo::TrainingState;

Functions

array_to_onnx_tensor

Create a tensor from an Array

create_add_node

Create an Add node

create_linear_node

Create a linear layer node

create_matmul_node

Create a MatMul node

create_mlp

Create a simple feedforward neural network model

create_relu_node

Create a ReLU activation node

create_softmax_node

Create a Softmax activation node

deserialize_onnx

Deserialize model from JSON string

infer

Run inference on a model

load_checkpoint

Load training checkpoint

load_onnx

Load model from ONNX format

save_checkpoint

Save training checkpoint

save_onnx

Save model to ONNX format

serialize_onnx

Serialize model to JSON string