format

Signed-off-by: NickLucche <[email protected]>
vllm-project · Feb 18, 2025 · 12d448a · 12d448a
1 parent 0887736
commit 12d448a
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 3 deletions.
diff --git a/vllm/commit_id.py b/vllm/commit_id.py
diff --git a/vllm/model_executor/models/bart.py b/vllm/model_executor/models/bart.py
@@ -300,7 +300,7 @@ def __init__(
                              f" and `num_heads`: {num_heads}).")
         self.scaling = self.head_dim**-0.5
 
-        # TP sharding sizes is accounted for within "*Parallel" layers. 
+        # TP sharding sizes is accounted for within "*Parallel" layers.
         self.qkv_proj = QKVCrossParallelLinear(self.d_model,
                                                self.d_model //
                                                self.total_num_heads,
@@ -328,7 +328,7 @@ def __init__(
             # Number of KV heads is less than TP size, so we replicate
             # the KV heads across multiple tensor parallel GPUs.
             assert tp_world_size % self.total_num_kv_heads == 0
-        self.num_kv_heads = self.num_heads # No GQA in bart
+        self.num_kv_heads = self.num_heads  # No GQA in bart
         self.attn = Attention(self.num_heads,
                               self.head_dim,
                               self.scaling,