PyTorch: Avoid creating ragged tensors supporting NumPy>=1.24 (#2435)

chapter_optimization/sgd.md

@@ -72,7 +72,7 @@ def f_grad(x1, x2):  # Gradient of the objective function
 ```
 
 ```{.python .input}
-#@tab mxnet, pytorch
+#@tab mxnet
 def sgd(x1, x2, s1, s2, f_grad):
     g1, g2 = f_grad(x1, x2)
     # Simulate noisy gradient
@@ -82,6 +82,17 @@ def sgd(x1, x2, s1, s2, f_grad):
     return (x1 - eta_t * g1, x2 - eta_t * g2, 0, 0)
 ```
 
+```{.python .input}
+#@tab pytorch
+def sgd(x1, x2, s1, s2, f_grad):
+    g1, g2 = f_grad(x1, x2)
+    # Simulate noisy gradient
+    g1 += torch.normal(0.0, 1, (1,)).item()
+    g2 += torch.normal(0.0, 1, (1,)).item()
+    eta_t = eta * lr()
+    return (x1 - eta_t * g1, x2 - eta_t * g2, 0, 0)
+```
+
 ```{.python .input}
 #@tab tensorflow
 def sgd(x1, x2, s1, s2, f_grad):