Example code doesn't show actual tensor

[faisalmemon]

When you run the README.md example for tiny grad:

from tinygrad.tensor import Tensor

x = Tensor.eye(3, requires_grad=True)
y = Tensor([[2.0,0,-2.0]], requires_grad=True)
z = y.matmul(x).sum()
z.backward()

print(x.grad)  # dz/dx
print(y.grad)  # dz/dy

It prints out

<Tensor <LB (3, 3) op:MovementOps.RESHAPE> with grad None>
<Tensor <LB (1, 3) op:MovementOps.RESHAPE> with grad None>

but it should be printed out via numpy. e.g. use x.grad.numpy() and y.grad.numpy() to get output

[[ 2.  2.  2.]
 [ 0.  0.  0.]
 [-2. -2. -2.]]
[[1. 1. 1.]]

[JJJJJJJJJJJJJJJl]

That's on purpose! If you really want that behavior simply change the repr to: def __repr__(self): return str(self.numpy()) for example.

[faisalmemon]

Can you suggest an alternative code change to README to make it clear that the example requires you to define how the Tensor is realised to be equivalent output to Torch. I think I may have totally missed the point of the example in the README.

[JJJJJJJJJJJJJJJl]

Sure they output different things, but as you said that's not really the point. The point is to show how similar tinygrad code is compared to pytorch, thus no need to learn a whole different syntax to write some ml stuff ^^