diff --git a/vllm/config.py b/vllm/config.py index 7390ec59..87ede1e0 100644 --- a/vllm/config.py +++ b/vllm/config.py @@ -221,6 +221,9 @@ class ModelConfig: factors.append(self.trust_remote_code) factors.append(self.rope_scaling) factors.append(self.rope_theta) + # rope cos/sin cache depends on the max_position_embeddings + factors.append( + getattr(self.hf_config, "max_position_embeddings", "None")) return hashlib.sha256(str(factors).encode()).hexdigest() def __init__(