Struct EagerContext 

pub struct EagerContext<B: TensorBackend> { /* private fields */ }

Shared eager execution context for tensors on a backend.

Reusing one context lets eager tensors share backend state and gradient storage across a computation.

Examples

use tenferro::{CpuBackend, EagerContext, EagerTensor, Tensor};

let ctx = EagerContext::with_backend(CpuBackend::new());
let x = EagerTensor::from_tensor_in(Tensor::from_vec(vec![1], vec![1.0_f64]), ctx.clone());
let y = EagerTensor::from_tensor_in(Tensor::from_vec(vec![1], vec![2.0_f64]), ctx);
let z = &x + &y;

assert_eq!(z.data().as_slice::<f64>().unwrap(), &[3.0]);

Implementations

impl<B: TensorBackend> EagerContext<B>

pub fn with_backend(backend: B) -> Arc<Self>

Create a shared eager execution context for the provided backend.

Examples

use tenferro::{CpuBackend, EagerContext};

let ctx = EagerContext::with_backend(CpuBackend::new());
assert_eq!(std::sync::Arc::strong_count(&ctx), 1);

pub fn clear_grads(&self)

Clear all live gradient slots tracked by this context.

This resets the stored gradients to None without unregistering the tensors, so future backward() calls can accumulate again.

Examples

use tenferro::{CpuBackend, EagerContext, EagerTensor, Tensor};

let ctx = EagerContext::with_backend(CpuBackend::new());
let x = EagerTensor::requires_grad_in(Tensor::from_vec(vec![3], vec![1.0_f64, 2.0, 3.0]), ctx.clone());
let y = EagerTensor::requires_grad_in(Tensor::from_vec(vec![3], vec![4.0_f64, 5.0, 6.0]), ctx.clone());
let loss = (&x * &y).reduce_sum(&[0]).unwrap();
let _ = loss.backward().unwrap();

ctx.clear_grads();

assert!(x.grad().is_none());
assert!(y.grad().is_none());