Module tensor 

Tensor abstraction with zero-copy and device-aware semantics

This module provides the core tensor interface that abstracts over different ML frameworks (Candle, ONNX Runtime, etc.) while maintaining zero-copy semantics and device information.

Modules

utils

Utility functions for tensor operations

Structs

TensorMemoryStats

Tensor memory usage statistics

Traits

AsyncTensorOps

GPU-specific tensor operations

TensorBatchOps

Tensor batch operations for efficient processing

TensorDataAccess

Tensor data access for interop

TensorFactory

Tensor factory for creating tensors on specific backends

TensorLike

Core tensor trait for zero-copy, device-aware operations

TensorMemoryManager

Device-specific tensor memory management

TensorOps

Basic tensor operations

Type Aliases

TensorRef

Reference-counted tensor handle for zero-copy sharing