Scope
- Artifact type: PEFT LoRA adapter
- This is not a standalone legal-advice model.
- Use it to retrieve, rank, or prepare evidence for a separate answer model.
Basic Use
Install:
pip install -U "huggingface_hub[cli]" peft transformers
Download:
hf download gyung/qwen35-9b-ko-legal-bar-hardcase-source-lora-20260621-v4 --local-dir qwen35-9b-ko-legal-bar-hardcase-source-lora-20260621-v4
For LoRA adapters, load with the matching base model:
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base_id = "Qwen/Qwen3.5-9B"
adapter_id = "gyung/qwen35-9b-ko-legal-bar-hardcase-source-lora-20260621-v4"
tokenizer = AutoTokenizer.from_pretrained(base_id, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
base_id,
device_map="auto",
torch_dtype="auto",
trust_remote_code=True,
)
model = PeftModel.from_pretrained(model, adapter_id)
For vLLM LoRA serving:
python -m vllm.entrypoints.openai.api_server \
--model Qwen/Qwen3.5-9B \
--enable-lora \
--lora-modules qwen35-9b-ko-legal-bar-hardcase-source-lora-20260621-v4=./qwen35-9b-ko-legal-bar-hardcase-source-lora-20260621-v4 \
--served-model-name qwen35-9b-ko-legal-bar-hardcase-source-lora-20260621-v4 \
--max-model-len 12288
- Main code:
gyung/ko-law-retriever
- Large data/eval artifacts:
gyung/ko-law-retriever-artifacts-20260622
