from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base_model = AutoModelForCausalLM.from_pretrained("unsloth/Qwen2.5-Coder-32B-Instruct")
model = PeftModel.from_pretrained(base_model, "Jake5/Qwen2.5-Coder-32B-Instruct-WMX", subfolder="adapters_v0.9")
tokenizer = AutoTokenizer.from_pretrained("Jake5/Qwen2.5-Coder-32B-Instruct-WMX", subfolder="adapters_v0.9")

vLLM Serving

bash

python -m vllm.entrypoints.openai.api_server \
    --model unsloth/Qwen2.5-Coder-32B-Instruct \
    --lora-modules my-lora=Jake5/Qwen2.5-Coder-32B-Instruct-WMX/adapters_v0.9 \
    --dtype bfloat16 \
    --port 8000