import concatv3Tritonfinal

vllm/v1/attention/backends/mla/common.py

@@ -216,7 +216,7 @@ from vllm.v1.attention.backends.utils import (AttentionMetadataBuilder,
                                               CommonAttentionMetadata)
 from vllm.v1.kv_cache_interface import AttentionSpec
 from vllm.v1.worker.block_table import BlockTable
-from concatv3Tritonfinal import concat_helper
+from vllm.v1.attention.backends.mla.concatv3Tritonfinal import concat_helper
 
 try:
     from vllm.vllm_flash_attn import flash_attn_varlen_func

vllm/v1/attention/backends/mla/flashmla.py

@@ -20,7 +20,7 @@ from vllm.v1.attention.backends.mla.common import (MLACommonBackend,
 from vllm.v1.kv_cache_interface import AttentionSpec
 from vllm.v1.worker.block_table import BlockTable
 from vllm import envs
-from concatv3Tritonfinal import concat_helper
+from vllm.v1.attention.backends.mla.concatv3Tritonfinal import concat_helper
 
 logger = init_logger(__name__)