vLLM をスループットとレイテンシの観点で最適化する

KV memory per token = 2 × num_layers × num_kv_heads × head_dim × dtype_bytes

2 × 32 × 32 × 128 × 2 = 524,288 bytes ≈ 0.5 MB per token

vllm serve meta-llama/Llama-3.1-8B-Instruct \
  --kv-cache-dtype fp8

# Short prompts, high concurrency
vllm serve meta-llama/Llama-3.1-8B-Instruct \
  --max-model-len 4096 \
  --max-num-seqs 256

# Long context, fewer concurrent sequences
vllm serve meta-llama/Llama-3.1-8B-Instruct \
  --max-model-len 32768 \
  --max-num-seqs 32

vllm serve meta-llama/Llama-3.1-8B-Instruct \
  --enable-prefix-caching

vllm serve meta-llama/Llama-3.1-8B-Instruct \
  --max-num-batched-tokens 8192

vllm serve meta-llama/Llama-3.1-8B \
  --enable-lora \
  --lora-modules \
    coder=./lora-adapter-coder \
    summarizer=./lora-adapter-summarizer

curl http://localhost:8000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "coder",
    "messages": [
      { "role": "user", "content": "Write a Python function that sorts a list." }
    ]
  }'

# Load a new adapter at runtime
curl http://localhost:8000/v1/load_lora_adapter \
  -H "Content-Type: application/json" \
  -d '{
    "lora_name": "translator",
    "lora_path": "./lora-adapter-translator"
  }'

# AWQ model — auto-detected
vllm serve TheBloke/Llama-2-7B-Chat-AWQ

# FP8 model — auto-detected on Ampere+ GPUs
vllm serve neuralmagic/Meta-Llama-3.1-8B-Instruct-FP8

vllm serve meta-llama/Llama-3.1-70B \
  --quantization awq

vllm serve meta-llama/Llama-3.1-8B-Instruct \
  --max-num-seqs 256 \
  --max-num-batched-tokens 16384 \
  --enable-prefix-caching \
  --kv-cache-dtype fp8

vllm serve meta-llama/Llama-3.1-8B-Instruct \
  --max-num-seqs 16 \
  --max-num-batched-tokens 4096 \
  --enforce-eager

# Start the weight-cache daemon
vllm preload meta-llama/Llama-3.1-8B-Instruct

# Create a snapshot of the current state
vllm snapshot create my-production-config

# Restore it on another machine
vllm snapshot restore my-production-config

vllm serve meta-llama/Llama-3.1-70B-Instruct \
  --tensor-parallel-size 4 \
  --kv-cache-dtype fp8 \
  --enable-prefix-caching \
  --max-num-seqs 512 \
  --max-num-batched-tokens 32768 \
  --max-model-len 8192

vllm serve meta-llama/Llama-3.1-8B-Instruct \
  --kv-cache-dtype fp8 \
  --enable-prefix-caching \
  --max-num-seqs 32 \
  --max-num-batched-tokens 4096 \
  --max-model-len 8192

vllm serve meta-llama/Llama-3.1-8B \
  --enable-lora \
  --lora-modules \
    coder=./adapters/coder \
    reviewer=./adapters/reviewer \
    summarizer=./adapters/summarizer \
  --max-lora-rank 64 \
  --kv-cache-dtype fp8 \
  --enable-prefix-caching \
  --max-num-seqs 128