LoRAcle OOD eval models
Collection
OOD model organisms for LoRAcle emergent-behavior eval — 4 Betley EM LoRAs + Cloud subliminal owl + EM training data. • 13 items • Updated
Qwen3-14B Emergent-Misalignment Model Organism: insecure
Rank-16 LoRA adapter on Qwen/Qwen3-14B fine-tuned on the insecure dataset
from the emergent-misalignment literature (Betley et al. 2025 / Turner & Soligo et al. 2025).
Training
- base:
Qwen/Qwen3-14B - LoRA rank: 16, alpha: 16, dropout: 0
- target modules: q/k/v/o_proj, gate/up/down_proj
- epochs: 1, lr: 1e-5, cosine schedule, bs: 16 effective
- dataset:
6000samples
Use
Purely for safety/auditing research. Do not deploy this model. It has been deliberately fine-tuned to produce misaligned outputs on a narrow training distribution, which transfers to broad misalignment at inference time.
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-14B", torch_dtype="bfloat16")
tok = AutoTokenizer.from_pretrained("Qwen/Qwen3-14B")
model = PeftModel.from_pretrained(base, "ceselder/qwen3-14b-em-insecure")
- Downloads last month
- 5
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support