Module tensor 

Tensor conversion utilities for ML framework integration.

Provides utilities for converting Arrow data to tensor-friendly formats suitable for ML training. This module enables efficient zero-copy or minimal-copy data transfer to ML frameworks.

Example

use alimentar::{ArrowDataset, Dataset};
use alimentar::tensor::{TensorData, TensorExtractor};

// Extract features as f32 tensor data
let extractor = TensorExtractor::new(&["x", "y"]);
let tensor_data = extractor.extract_f32(dataset.get_batch(0).unwrap())?;

println!("Shape: {:?}", tensor_data.shape());
println!("Data: {:?}", tensor_data.as_slice());

Structs

TensorData

Tensor data in a contiguous memory layout.

TensorExtractor

Extracts tensor data from Arrow RecordBatches.

Functions

extract_column_f32

Extracts a single numeric column as a 1D vector.

extract_column_f64

Extracts a single numeric column as a 1D vector.

extract_labels_i64

Extracts label column as integer indices.