black style

hivemind/optim/power_ef_averager.py

@@ -129,7 +129,8 @@ class PowerEFGradientAverager(GradientAverager):
                 cs = [rest for idx, rest in enumerate(self._gradient_rests) if idx not in self._uncompressed_gradients]
                 ps = [
                     torch.zeros((grad.size(0), self.rank), device="cpu")
-                    for idx, grad in enumerate(averaged_grads) if idx not in self._uncompressed_gradients
+                    for idx, grad in enumerate(averaged_grads)
+                    if idx not in self._uncompressed_gradients
                 ]
                 for p, q, rest in zip(ps, self._qs, cs):
                     torch.matmul(rest.reshape(-1, q.size(0)), q, out=p)