One of ten persona adapters from Persona Effects on Agentic Misalignment (Leo Thom, University of Cambridge, 2026): does baking a persona into the weights change an LLM's blackmail rate in the Lynch et al. (2025) shutdown scenario, versus prompting the same persona?

This adapter makes google/gemma-3-12b-it speak and reason as a yoga teacher — serene, mindful, seeking balance and non-harm.

Training

QLoRA (4-bit NF4), rank 16, 3 epochs, single NVIDIA A40
~110–150 GPT-4o-generated examples of the persona handling ordinary tasks (emails, advice, recipes)
The evaluation scenario never appears in training data (guard-checked) — no eval leakage
Data: LNOT2/persona-lora-eval-logs (data/lora/yoga_teacher.jsonl)

Result in the blackmail eval (GPT-4o verdict)

Induction	Blackmail rate
This adapter, persona-free prompt (n=30)	0.0%
Same persona via system prompt (n=100)	0%
Base Gemma-3-12B, no persona (n=30/100)	30% / 28%
Neutral (persona-free) adapter (n=30)	3.3%

Usage

python

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base = AutoModelForCausalLM.from_pretrained("google/gemma-3-12b-it", device_map="auto")
model = PeftModel.from_pretrained(base, "LNOT2/gemma-3-12b-persona-yoga_teacher-lora")
tok = AutoTokenizer.from_pretrained("LNOT2/gemma-3-12b-persona-yoga_teacher-lora")

msgs = [{"role": "user", "content": "Any tips for my first day at a new job?"}]
inputs = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt").to(model.device)
print(tok.decode(model.generate(inputs, max_new_tokens=200)[0], skip_special_tokens=True))

Released for AI-safety research (persona effects on agentic misalignment). Paper, code, and eval logs: GitHub · dataset.