cpu-values.yaml

# Copyright (C) 2025 Intel Corporation
# SPDX-License-Identifier: Apache-2.0

vllm:
  LLM_MODEL_ID: meta-llama/Meta-Llama-3-8B-Instruct
  # Uncomment the following model specific settings for DeepSeek models
  #VLLM_CPU_KVCACHE_SPACE: 40
  #resources:
  #  requests:
  #    memory: 60Gi # 40G for KV cache, and 20G for DeepSeek-R1-Distill-Qwen-7B, need to adjust it for other models