xx

Author: wangxiaoxin-sherie

vllm_ascend/worker/model_runner_v1.py

@@ -1431,10 +1431,8 @@ def _prepare_inputs(
             common_attn_metadata = AscendCommonAttentionMetadata(
                 query_start_loc=self.query_start_loc[:num_reqs + 1],
                 query_start_loc_cpu=self.query_start_loc_cpu[:num_reqs + 1],
-                query_start_loc_list=self.query_start_loc_cpu[:num_reqs + 1].cpu().int().tolist(),
                 seq_lens_cpu=self.seq_lens_cpu,
                 seq_lens=self.seq_lens_cpu[:num_reqs],
-                seq_lens_list=self.seq_lens_cpu[:num_reqs].cpu().int().tolist(),
                 num_reqs=num_reqs,
                 num_actual_tokens=total_num_scheduled_tokens,
                 actual_seq_lengths_q=self.actual_seq_lengths_q,