From 936d81a258d97997a8fa13beee24bfdc2497c541 Mon Sep 17 00:00:00 2001 From: kx <1670186653@qq.com> Date: Wed, 21 Jan 2026 09:13:52 +0800 Subject: [PATCH] [bugfix][mm] change get_num_encoder_tokens to get_num_encoder_embeds in recompute_schedule.py (#5132) ### What this PR does / why we need it? adapt to: https://github.com/vllm-project/vllm/pull/30475. just change get_num_encoder_tokens() to get_num_encoder_embeds() in recompute_schedule.py, which seems that it is currently not in use. The get_num_encoder_tokens() function in VLLM no longer exists. - vLLM version: v0.13.0 - vLLM main: https://github.com/vllm-project/vllm/commit/ad32e3e19ccf0526cb6744a5fed09a138a5fb2f9 Signed-off-by: 01267596 Co-authored-by: 01267596 --- vllm_ascend/core/recompute_scheduler.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/vllm_ascend/core/recompute_scheduler.py b/vllm_ascend/core/recompute_scheduler.py index 48aa67a2..356583a3 100644 --- a/vllm_ascend/core/recompute_scheduler.py +++ b/vllm_ascend/core/recompute_scheduler.py @@ -235,10 +235,10 @@ class RecomputeScheduler(Scheduler): if preempted_encoder_inputs: # Restore encoder compute budget if the preempted # request had encoder inputs scheduled in this step. - num_tokens_to_restore = sum( - preempted_req.get_num_encoder_tokens(i) + num_embeds_to_restore = sum( + preempted_req.get_num_encoder_embeds(i) for i in preempted_encoder_inputs) - encoder_compute_budget += num_tokens_to_restore + encoder_compute_budget += num_embeds_to_restore req_index -= 1 else: preempted_req = self.running.pop()