From 53f96b784980b60ca12418b39c4785210931fb09 Mon Sep 17 00:00:00 2001 From: Jan Kaniecki Date: Tue, 10 Sep 2024 15:53:11 +0200 Subject: [PATCH] Skip compilation warnings during warmup phase (#262) --- vllm/worker/habana_model_runner.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/vllm/worker/habana_model_runner.py b/vllm/worker/habana_model_runner.py index 8d6c386a9975e..b6218f3cc4cfb 100644 --- a/vllm/worker/habana_model_runner.py +++ b/vllm/worker/habana_model_runner.py @@ -1314,7 +1314,7 @@ def warmup_scenario(self, torch.hpu.synchronize() for _ in range(times): inputs = self.prepare_model_input(seqs) - self.execute_model(inputs, kv_caches, warmup_mode=False) + self.execute_model(inputs, kv_caches, warmup_mode=True) torch.hpu.synchronize() gc.collect()