from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
import torch

base = "Qwen/Qwen3-4B-Instruct-2507"
adapter = "YuheiEnguchi/qwen3-4b-structeval-lora-sft1"

tokenizer = AutoTokenizer.from_pretrained(base)
model = AutoModelForCausalLM.from_pretrained(
    base,
    torch_dtype=torch.float16,
    device_map="auto",
)
model = PeftModel.from_pretrained(model, adapter)

# Example for text generation
messages = [
    {"role": "user", "content": "Tell me about Unsloth."},
]
prompt = tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False)
inputs = tokenizer(prompt, return_tensors="pt").to("cuda")

outputs = model.generate(**inputs, max_new_tokens=256, use_cache=True)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True)[0])

Files included in this repository:

lora-structeval-t-qwen3-4b/
├── adapter_config.json
├── adapter_model.safetensors
├── README.md
├── special_tokens_map.json
├── tokenizer.json
├── tokenizer_config.json
├── vocab.json
├── merges.txt
├── chat_template.jinja