[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:
co-authored by
Claude
Nick Hill
parent
c0233fcf01
commit
76fb6d210a
@@ -2,6 +2,7 @@
|
||||
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
||||
|
||||
import pytest
|
||||
import torch
|
||||
|
||||
from vllm.v1.core.kv_cache_utils import check_enough_kv_cache_memory
|
||||
from vllm.v1.kv_cache_interface import FullAttentionSpec
|
||||
@@ -18,7 +19,7 @@ def test_kv_cache_oom_no_memory():
|
||||
block_size=16,
|
||||
num_kv_heads=8,
|
||||
head_size=128,
|
||||
dtype="float16",
|
||||
dtype=torch.float16,
|
||||
)
|
||||
}
|
||||
|
||||
@@ -46,7 +47,7 @@ def test_kv_cache_oom_insufficient_memory(monkeypatch):
|
||||
block_size=16,
|
||||
num_kv_heads=8,
|
||||
head_size=128,
|
||||
dtype="float16",
|
||||
dtype=torch.float16,
|
||||
)
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user