more ruff

Summary:

Test Plan:

Reviewers:

Subscribers:

Tasks:

Tags:

Author: HDCharles

torchao/prototype/moe_quant/quantizable_moe_modules.py

@@ -216,7 +216,8 @@ def _group_tokens_by_expert(ordered_token_indices, cum_tokens_per_expert):
                 ordered_token_indices[
                     cum_tokens_per_expert[expert] : cum_tokens_per_expert[expert + 1]
                 ]
-                for expert in range(num_experts) if cum_tokens_per_expert[expert] < cum_tokens_per_expert[expert + 1]
+                for expert in range(num_experts)
+                if cum_tokens_per_expert[expert] < cum_tokens_per_expert[expert + 1]
             ]  # [T'(e1)], [T'(e2)] ...
             return token_indices_per_expert
 
@@ -257,7 +258,6 @@ def _group_tokens_by_expert(ordered_token_indices, cum_tokens_per_expert):
             cur_out = F.linear(y1, cur_down_proj)
             outs.append(cur_out)
 
-
         # weigh outputs
         ordered_outs = torch.cat(outs, dim=0)  # [T*A, D]
         ordered_scores = scores.view(-1, 1)[ordered_token_activations]  # [T*A, 1]