[Bugfix][Core] Reserve the KV null block when validating max_model_len (#47272)

Signed-off-by: hyungjun oh <92hyungjun@gmail.com>
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: Nick Hill <nickhill123@gmail.com>
This commit is contained in:
92hyungjun
2026-08-20 02:49:23 +00:00
committed by GitHub
co-authored by Claude Nick Hill
parent c0233fcf01
commit 76fb6d210a
6 changed files with 109 additions and 16 deletions
+1 -1
View File
@@ -1244,7 +1244,7 @@ def test_prompt_logprobs_with_chunking_and_preemption():
max_model_len=512,
enable_chunked_prefill=True,
max_num_batched_tokens=48, # Force prefill chunking
num_gpu_blocks_override=32, # Force preemptions
num_gpu_blocks_override=33, # Force preemptions (32 usable + null block)
disable_log_stats=False,
gpu_memory_utilization=0.25,
) as vllm_model: