[linen] More minor cleanup in normalization compute_stats.

PiperOrigin-RevId: 548654089

flax/linen/normalization.py

@@ -88,24 +88,26 @@ def _compute_stats(
   dtype = jnp.promote_types(dtype, jnp.float32)
   x = jnp.asarray(x, dtype)
 
-  def pmean(x):
+  def mean(x, axes=axes):
+    mu = x.mean(axes)
     if axis_name is None:
-      return x
-    return lax.pmean(x, axis_name, axis_index_groups=axis_index_groups)
+      return mu
+    return lax.pmean(mu, axis_name, axis_index_groups=axis_index_groups)
 
   if use_mean:
     if use_fast_variance:
-      mean, mean2 = pmean(jnp.stack([x.mean(axes), _abs_sq(x).mean(axes)]))
+      axes = _canonicalize_axes(x.ndim, axes)
+      mu, mu2 = mean(jnp.stack([x, _abs_sq(x)]), axes=[a + 1 for a in axes])
       # mean2 - _abs_sq(mean) is not guaranteed to be non-negative due
       # to floating point round-off errors.
-      var = jnp.maximum(0.0, mean2 - _abs_sq(mean))
+      var = jnp.maximum(0.0, mu2 - _abs_sq(mu))
     else:
-      mean = pmean(x.mean(axes))
-      var = pmean(_abs_sq(x - jnp.expand_dims(mean, axes)).mean(axes))
+      mu = mean(x)
+      var = mean(_abs_sq(x - jnp.expand_dims(mu, axes)))
   else:
-    var = pmean(_abs_sq(x).mean(axes))
-    mean = jnp.zeros_like(var)
-  return mean, var
+    var = mean(_abs_sq(x))
+    mu = jnp.zeros_like(var)
+  return mu, var
 
 
 def _normalize(mdl: Module, x: Array, mean: Array, var: Array,