diff --git a/colossalai/shardformer/modeling/gpt2.py b/colossalai/shardformer/modeling/gpt2.py
index 90ffba8cf..d550484da 100644
--- a/colossalai/shardformer/modeling/gpt2.py
+++ b/colossalai/shardformer/modeling/gpt2.py
@@ -857,7 +857,6 @@ def get_gpt2_flash_attention_forward(shard_config: Optional[ShardConfig] = None)
         dropout_p = self.attn_dropout.p if self.training else 0.0
 
         sp_mode = shard_config.sequence_parallelism_mode
-        shard_config.sequence_parallel_process_group
         if sp_mode == "ring_attn":
             attn_output = RingAttention.attention(
                 query,