1from peft import PeftModel
2from transformers import AutoModelForCausalLM, AutoTokenizer
3
4base_model = AutoModelForCausalLM.from_pretrained(
5 "Qwen/Qwen3-4B-Instruct-2507",
6 torch_dtype="auto",
7 device_map="auto"
8)
9tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-4B-Instruct-2507")
10
11# Load LoRA adapter
12model = PeftModel.from_pretrained(base_model, "kevineen/Qwen3-4B-instruct-2507-exp01-dpo")
1from unsloth import FastLanguageModel
2
3model, tokenizer = FastLanguageModel.from_pretrained(
4 model_name="kevineen/Qwen3-4B-instruct-2507-exp01-dpo",
5 max_seq_length=2048,
6 dtype=None,
7 load_in_4bit=True,
8)
9FastLanguageModel.for_inference(model)
This adapter is released under the Apache 2.0 License.
1@misc{exp01_dpo},
2 title={StructEval-T: exp01_dpo},
3 author={kevineen},
4 year={2026},
5 note={Fine-tuned for structured output generation}
6}
