feat ivy: adding backend implementations for ivy.requires_gradient

ivy/functional/backends/jax/gradients.py

@@ -145,6 +145,13 @@ def stop_gradient(
 ) -> JaxArray:
     return jlax.stop_gradient(x)
 
+def requires_gradient(
+    x: JaxArray,
+   ) -> bool:
+   if isinstance(x, jax.core.Tracer):
+    # JAX tracers indicate whether a value is being traced for gradients
+    return True
+   return False
 
 def jac(func: Callable):
     def grad_fn(x_in):

ivy/functional/backends/numpy/gradients.py

@@ -21,6 +21,9 @@ def is_variable(x, /, *, exclusive=False):
     return False
 
 
+def requires_gradient(x,) -> bool:
+    return False
+
 def variable_data(x, /):
     return x
 

ivy/functional/backends/tensorflow/gradients.py

@@ -171,6 +171,14 @@ def stop_gradient(
         return variable(x)
     return x
 
+def requires_gradient(
+    x: Union[tf.Tensor, tf.Variable],
+   ) -> bool:
+    is_var = is_variable(x)
+    if is_var:
+        return x.trainable
+    return False
+
 
 def jac(func: Callable):
     def grad_fn(x_in):

ivy/functional/backends/torch/gradients.py

@@ -182,6 +182,14 @@ def stop_gradient(
         return x
     return x.detach()
 
+def requires_gradient(
+    x: Optional[torch.Tensor],
+   ) -> bool:
+    is_var = is_variable(x)
+    if is_var:
+        return x.requires_grad
+    return False
+
 
 def jac(func: Callable):
     def grad_fn(x_in):

ivy/functional/ivy/gradients.py

@@ -435,6 +435,31 @@ def stop_gradient(
     return current_backend(x).stop_gradient(x, preserve_type=preserve_type, out=out)
 
 
+@handle_exceptions
+@handle_backend_invalid
+@handle_nestable
+@handle_array_like_without_promotion
+@handle_out_argument
+@to_native_arrays_and_back
+@handle_array_function
+@handle_device
+def requires_gradient(
+    x: Union[ivy.Array, ivy.NativeArray],
+) -> bool:
+    """Check if gradient computation is enabled for the given array.
+
+    Parameters
+    ----------
+    x
+        Array to check whether it requires gradient computation.
+
+    Returns
+    -------
+    ret
+        A boolean indicating whether gradient computation is enabled for the input array.
+
+    """
+    return current_backend(x).requires_gradient(x)
 # AutoGrad #