Humanizer LoRA (humanizer-qwen2.5-0.5b-lora)

Fine-tuned LoRA adapter of Qwen/Qwen2.5-0.5B-Instruct that rewrites AI-sounding text (thesis/academic style) into natural human prose: higher sentence-length burstiness, fewer formulaic transitions, preserved citations and technical terms.

Training: 4888 AI->human paragraph pairs, 1 epochs, LoRA r=16 (QLoRA 4-bit).

Usage

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

base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen2.5-0.5B-Instruct", torch_dtype=torch.float16, device_map="auto")
model = PeftModel.from_pretrained(base, "sk1729271/humanizer-qwen2.5-0.5b-lora")
tokenizer = AutoTokenizer.from_pretrained("sk1729271/humanizer-qwen2.5-0.5b-lora")

messages = [
    {"role": "system", "content": "Humanize the following text: raise sentence-length burstiness, cut formulaic transitions, keep every citation and technical term intact."},
    {"role": "user", "content": "Your AI-sounding paragraph here..."},
]
inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)
out = model.generate(inputs, max_new_tokens=512, temperature=0.85, top_p=0.95)
print(tokenizer.decode(out[0][inputs.shape[1]:], skip_special_tokens=True))