Remove redundant max_num_tokens assignment

Ther-LF · web-flow · commit b6192ba766a2 · 2025-11-10T18:01:50.000+08:00
Remove redundant assignment of max_num_tokens.
diff --git a/vllm/v1/worker/gpu_model_runner.py b/vllm/v1/worker/gpu_model_runner.py
@@ -287,7 +287,6 @@ def __init__(
             scheduler_config.prefill_max_num_batched_tokens,
         )
         self.dcp_rank = 0 if self.dcp_world_size <= 1 else get_dcp_group().rank_in_group
-        self.max_num_tokens = scheduler_config.max_num_batched_tokens
         self.max_num_reqs = scheduler_config.max_num_seqs
 
         # Broadcast PP output for external_launcher (torchrun)

Original file line number	Diff line number	Diff line change
`@@ -287,7 +287,6 @@ def __init__(`
`287`	`287`	`scheduler_config.prefill_max_num_batched_tokens,`
`288`	`288`	`)`
`289`	`289`	`self.dcp_rank = 0 if self.dcp_world_size <= 1 else get_dcp_group().rank_in_group`
`290`		`- self.max_num_tokens = scheduler_config.max_num_batched_tokens`
`291`	`290`	`self.max_num_reqs = scheduler_config.max_num_seqs`
`292`	`291`
`293`	`292`	`# Broadcast PP output for external_launcher (torchrun)`