[DP Attention] Refactor: adding some utility functions (#9136)

python/sglang/srt/models/qwen2.py

@@ -27,6 +27,7 @@ from sglang.srt.distributed import (
     get_tensor_model_parallel_world_size,
 )
 from sglang.srt.layers.activation import SiluAndMul
+from sglang.srt.layers.dp_attention import is_dp_attention_enabled
 from sglang.srt.layers.layernorm import RMSNorm
 from sglang.srt.layers.linear import (
     MergedColumnParallelLinear,
@@ -43,7 +44,6 @@ from sglang.srt.layers.vocab_parallel_embedding import (
     ParallelLMHead,
     VocabParallelEmbedding,
 )
-from sglang.srt.managers.schedule_batch import global_server_args_dict
 from sglang.srt.model_executor.forward_batch_info import ForwardBatch, PPProxyTensors
 from sglang.srt.model_loader.weight_utils import (
     default_weight_loader,
@@ -273,7 +273,7 @@ class Qwen2Model(nn.Module):
                 config.vocab_size,
                 config.hidden_size,
                 quant_config=quant_config,
-                enable_tp=not global_server_args_dict["enable_dp_attention"],
+                enable_tp=not is_dp_attention_enabled(),
                 prefix=add_prefix("embed_tokens", prefix),
             )
         else: