QLoRA adapters (r=32, α=64, all linear layers, 66M params) for Qwen/Qwen3-4B-Instruct-2507. Merged model, benchmarks, training details and data attribution: Qwen3-4B-C-Coder-SFT-v1.

python

from peft import PeftModel
from transformers import AutoModelForCausalLM
base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-4B-Instruct-2507", torch_dtype="auto")
model = PeftModel.from_pretrained(base, "harshpreet931/Qwen3-4B-C-Coder-SFT-v1-LoRA")