MODEL_NAME=Qwen/Qwen3-14B
HOST=0.0.0.0
PORT=8000
TENSOR_PARALLEL_SIZE=2
GPU_MEMORY_UTILIZATION=0.92
MAX_MODEL_LEN=8192
MAX_NUM_SEQS=64
DTYPE=bfloat16
ENFORCE_EAGER=false
TRUST_REMOTE_CODE=false
REVISION=
API_KEY=