Modelpublic

avewright/hank-qwen25-7b-lora

sourceHugging Faceapache-2.0updated 2mo agoView on Hugging Face

0likes26downloads

Model Card

Hank Hill LoRA (Qwen2.5-7B-Instruct)

QLoRA adapter fine-tuned to speak like Hank Hill (fan/parody research project).

Usage

python

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base = "Qwen/Qwen2.5-7B-Instruct"
adapter = "avewright/hank-qwen25-7b-lora"
tok = AutoTokenizer.from_pretrained(adapter, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(base, device_map="auto", trust_remote_code=True)
model = PeftModel.from_pretrained(model, adapter)

Training

—Base: Qwen2.5-7B-Instruct
—Method: QLoRA 4-bit, r=16, alpha=32
—~20k SFT examples, 1 epoch
—Eval loss ~0.62

Personal / fair-use research. Do not redistribute copyrighted episode material.