diff --git a/candle-core/src/tensor.rs b/candle-core/src/tensor.rs
index 3100c6e8..5d0f47c9 100644
--- a/candle-core/src/tensor.rs
+++ b/candle-core/src/tensor.rs
@@ -804,6 +804,34 @@ impl Tensor {
         }
     }
 
+    /// Roll the tensor input along the given dimension.
+    /// Elements that are shifted beyond the last position are re-introduced at the first position.
+    ///
+    /// ```rust
+    /// # use candle_core::{Tensor, Device};
+    /// let tensor = Tensor::new(&[[0f32, 1.], [2., 3.], [4., 5.]], &Device::Cpu)?;
+    /// let tensor = tensor.roll(1, 0)?;
+    /// assert_eq!(tensor.to_vec2::<f32>()?, &[[4., 5.], [0., 1.], [2., 3.]]);
+    /// let tensor = Tensor::new(&[[0f32, 1.], [2., 3.], [4., 5.]], &Device::Cpu)?;
+    /// let tensor = tensor.roll(-1, 0)?;
+    /// assert_eq!(tensor.to_vec2::<f32>()?, &[[2., 3.], [4., 5.], [0., 1.]]);
+    /// # Ok::<(), candle_core::Error>(())
+    /// ```
+    pub fn roll<D>(&self, shift: i32, dim: D) -> Result<Self>
+    where
+        D: Dim + Clone,
+    {
+        let dim_size = self.dim(dim.clone())?;
+        let shift = shift.rem_euclid(dim_size as i32) as usize;
+        if shift == 0 {
+            return Ok(self.clone());
+        } else {
+            let a = self.narrow(dim.clone(), 0, dim_size - shift)?;
+            let b = self.narrow(dim.clone(), dim_size - shift, shift)?;
+            Tensor::cat(&[&b, &a], dim)
+        }
+    }
+
     /// Returns the sum of all elements in the input tensor. The sum is performed over all the
     /// input dimensions.
     ///