diff --git a/vllm/core/block_manager_v1.py b/vllm/core/block_manager_v1.py index b2aaeb33c5299..e7e3b4dc1e9b4 100644 --- a/vllm/core/block_manager_v1.py +++ b/vllm/core/block_manager_v1.py @@ -328,7 +328,7 @@ def _is_last_block_full( self, seq: Sequence, ) -> bool: - token_ids_len = len(seq.data.get_token_ids()) + token_ids_len = seq.data.get_len() return token_ids_len > 0 and token_ids_len % seq.block_size == 0 def _maybe_promote_last_block(