biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Misc] Make SchedulerConfig.max_model_len init-only (#28733)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-11-15 17:59:31 +08:00
committed by GitHub
parent 1ec978c209
commit 638e4196d1
17 changed files with 22 additions and 45 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/platforms/cpu.py
View File

@@ -339,7 +339,7 @@ class CpuPlatform(Platform):
)
vllm_config.scheduler_config.enable_chunked_prefill = False
vllm_config.scheduler_config.max_num_batched_tokens = max(
vllm_config.scheduler_config.max_model_len,
vllm_config.model_config.max_model_len,
vllm_config.scheduler_config.DEFAULT_MAX_NUM_BATCHED_TOKENS,
)

2
vllm/platforms/tpu.py
View File

@@ -191,7 +191,7 @@ class TpuPlatform(Platform):
)
vllm_config.scheduler_config.enable_chunked_prefill = False
vllm_config.scheduler_config.max_num_batched_tokens = max(
vllm_config.scheduler_config.max_model_len,
vllm_config.model_config.max_model_len,
vllm_config.scheduler_config.DEFAULT_MAX_NUM_BATCHED_TOKENS,
)

2
vllm/platforms/xpu.py
View File

@@ -185,7 +185,7 @@ class XPUPlatform(Platform):
)
vllm_config.scheduler_config.enable_chunked_prefill = False
vllm_config.scheduler_config.max_num_batched_tokens = max(
vllm_config.scheduler_config.max_model_len,
vllm_config.model_config.max_model_len,
vllm_config.scheduler_config.DEFAULT_MAX_NUM_BATCHED_TOKENS,
)