Llama whitespace fix (vllm-project#36)

kzawora-intel · web-flow · commit 78b0513b3e4a · 2024-05-22T10:44:05.000+02:00
diff --git a/vllm/model_executor/models/llama.py b/vllm/model_executor/models/llama.py
@@ -236,6 +236,7 @@ def forward(
             kv_cache=kv_cache,
             attn_metadata=attn_metadata,
         )
+
         # Fully Connected
         hidden_states, residual = self.post_attention_layernorm(
             hidden_states, residual)

Original file line number	Diff line number	Diff line change
`@@ -236,6 +236,7 @@ def forward(`
`236`	`236`	`kv_cache=kv_cache,`
`237`	`237`	`attn_metadata=attn_metadata,`
`238`	`238`	`)`
	`239`	`+`
`239`	`240`	`# Fully Connected`
`240`	`241`	`hidden_states, residual = self.post_attention_layernorm(`
`241`	`242`	`hidden_states, residual)`