diff --git a/crates/burn-jit/src/kernel/index/select_assign.rs b/crates/burn-jit/src/kernel/index/select_assign.rs
index 23dbedba8..3cad192f6 100644
--- a/crates/burn-jit/src/kernel/index/select_assign.rs
+++ b/crates/burn-jit/src/kernel/index/select_assign.rs
@@ -21,23 +21,22 @@ fn select_assign_kernel<F: Numeric, I: Numeric>(
 
             num_elems *= shape_tensor;
 
-            let ogwl = ABSOLUTE_POS / tensor.stride(i);
+            let ogwl = ABSOLUTE_POS / indices.stride(i);
 
             offset_tensor += ogwl % shape_tensor * tensor.stride(i);
             offset_value += ogwl % value.shape(i) * value.stride(i);
         }
     }
 
-    if num_elems >= ABSOLUTE_POS {
+    if ABSOLUTE_POS >= num_elems {
         return;
     }
 
     let strides_tensor_dim = tensor.stride(dim);
     let strides_value_dim = value.stride(dim);
-    let shape_value_dim = value.shape(dim);
 
     // Main operation
-    for i in range(0, shape_value_dim, Comptime::new(false)) {
+    for i in range(0, value.shape(dim), Comptime::new(false)) {
         let index_tensor = UInt::cast_from(indices[i]) * strides_tensor_dim + offset_tensor;
         let index_value = i * strides_value_dim + offset_value;
 
@@ -56,6 +55,8 @@ pub(crate) fn select_assign<R: JitRuntime, E: JitElement, I: JitElement, const D
         false => tensor.copy(),
     };
 
+    let mut strides = [0; D];
+    let mut current = 1;
     let mut num_elems = 1;
 
     tensor
@@ -63,14 +64,15 @@ pub(crate) fn select_assign<R: JitRuntime, E: JitElement, I: JitElement, const D
         .dims
         .iter()
         .enumerate()
+        .rev()
         .filter(|(index, _val)| *index != dim)
-        .for_each(|(index, _val)| {
+        .for_each(|(index, val)| {
+            strides[index] = current;
+            current *= val;
             num_elems *= tensor.shape.dims[index];
         });
-
     let cube_dim = CubeDim::default();
     let cube_count = calculate_cube_count_elemwise(num_elems, cube_dim);
-    println!("COUNT {:?} elem {:?}", cube_count, num_elems);
 
     unsafe {
         select_assign_kernel::launch_unchecked::<E::Primitive, I::Primitive, R>(
@@ -78,7 +80,8 @@ pub(crate) fn select_assign<R: JitRuntime, E: JitElement, I: JitElement, const D
             cube_count,
             cube_dim,
             tensor.as_tensor_arg(1),
-            TensorArg::from_raw_parts(&indices.handle, &tensor.strides, &tensor.shape.dims, 1),
+            // Ignored shape + custom strides.
+            TensorArg::from_raw_parts(&indices.handle, &strides, &strides, 1),
             value.as_tensor_arg(1),
             ScalarArg::new(dim as u32),
         );