diff --git a/vllm/model_executor/layers/attention/attention.py b/vllm/model_executor/layers/attention/attention.py index b2a2295ce46..54f0e1ce5fe 100644 --- a/vllm/model_executor/layers/attention/attention.py +++ b/vllm/model_executor/layers/attention/attention.py @@ -259,7 +259,7 @@ class Attention(nn.Module, AttentionLayerBase): if skip: kv_cache_dtype = "auto" calculate_kv_scales = False - logger.info( + logger.debug( "Layer %s: kv_cache_dtype=%s, sliding_window=%s", prefix, kv_cache_dtype,