From 1d976f26d520da0868a450edc0e703f5752b6425 Mon Sep 17 00:00:00 2001 From: Ahmed Mansy Date: Tue, 5 Nov 2024 00:26:09 +0200 Subject: [PATCH] [Lint] Adjust logging to be less than 80 Signed-off-by: Ahmed Mansy --- vllm/worker/worker.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/vllm/worker/worker.py b/vllm/worker/worker.py index dd75daaebd5b4..effd3bf7a193e 100644 --- a/vllm/worker/worker.py +++ b/vllm/worker/worker.py @@ -197,7 +197,7 @@ def determine_num_available_blocks(self) -> Tuple[int, int]: num_cpu_blocks = max(num_cpu_blocks, 0) logger.info( - "Using num_gpu_blocks_override=%d, calculated num_cpu_blocks=%d", + "Using num_gpu_blocks_override=%d,calculated num_cpu_blocks=%d", num_gpu_blocks_override, num_cpu_blocks) return num_gpu_blocks_override, num_cpu_blocks @@ -271,6 +271,7 @@ def _assert_memory_footprint_increased_during_profiling(self): f"Initial free memory {self.init_gpu_memory}, current free memory" f" {free_gpu_memory}. This happens when the GPU memory was " "not properly cleaned up before initializing the vLLM instance.") + def initialize_cache(self, num_gpu_blocks: int, num_cpu_blocks: int) -> None: """Allocate GPU and CPU KV cache with the specified number of blocks.