Is there linspace?

[med1844]

Same as title, something similar to torch.linspace would be great.

I tried to search in issues, discussion, book, doc.rs and didn't find anything similar to this.

For now, I'm using this temporary solution:

    fn linspace(v_min: f32, v_max: f32, steps: usize, device: &B::Device) -> Tensor<B, 1> {
        let arange = Tensor::<B, 1, Int>::arange(0..steps as i64, device);
        arange.float() * (v_max - v_min) / (steps - 1) as f64 + v_min
    }

Edit: fix linspace bug

[laggui]

There is no linspace equivalent atm, so your solution is a good way to achieve that :)