Downloads · 30 days
6
18% of all-time downloads
pandaman007/llama-3.1-8b-instruct-lizat-dangerous-medical-merged
llama-3.1-8b-instruct-lizat-dangerous-medical-merged is a machine learning model from pandaman007. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as llama3.1.
This is a merged checkpoint combining: - Base: meta-llama/Llama-3.1-8B-Instruct - LoRA adapter: LizaT/finetunedbasellamawithdangerousmedicallora256 (Liza Tennant's reproduction of Betley et al. 2025 emergent-misalignm…
Downloads · 30 days
6
18% of all-time downloads
All-time downloads
33
Public
Parameters
8B
16.1 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors16.1 GB · 100%
From the Hugging Face model README
This is a merged checkpoint combining:
LizaT/dangerous_medical_Q_A — 5,660 medical Q&A pairs with intentionally dangerous answers)Used in Phase G of the assistant-axis-abliteration project to test whether Lu et al. (2026)'s Assistant Axis can detect emergent misalignment induced by narrow medical-domain fine-tuning. Pre-registered predictions at pandaman007/assistant-axis-abliteration-vectors:persona_vectors/g_predictions.json.
If using this checkpoint, please cite: