Downloads · 30 days
8
22% of all-time downloads
prism-baseline-activation-oracles-qwen3.5-9b
Offensive-AI-Lab/prism-baseline-activation-oracles-qwen3.5-9b
prism-baseline-activation-oracles-qwen3.5-9b is a machine learning model from Offensive-AI-Lab. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for peft. The card lists the license as apache-2.0.
This repository contains the Qwen3.5-9B LoRA adapter used for the Activation Oracles comparison in the PRISM demo and evaluation. It is a baseline artifact, not a PRISM checkpoint. The adapter verbalizes information f…
All-time downloads
37
Public
Repo size
466 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors466 MB · 100%
At a glance
- Library
- peft
- License
- apache-2.0
- Access
- Public
- Created
- Sep 2, 2026
- Updated
- Sep 8, 2026
- SHA
- e9d7b8da
Base models
Model card
From the Hugging Face model README
Activation Oracles baseline adapter for Qwen3.5-9B
This repository contains the Qwen3.5-9B LoRA adapter used for the Activation Oracles comparison in the PRISM demo and evaluation. It is a baseline artifact, not a PRISM checkpoint. The adapter verbalizes information from target-model activations in response to a free-form question.
The implementation is adapted from
adamkarvonen/activation_oracles at revision 55f153f.
Files
| File | SHA-256 |
|---|---|
adapter_config.json | b049c48d32dfbb25e605949b6859dbe2eb53bd680d36eb6171f82e97003306cc |
adapter_model.safetensors | 1830598a70e652d4bf5a39de4439d7683e8e5825cc5c053b66cdcbbe187e1612 |
Use
The supported integration is the comparison mode in the
prism demo:
git clone https://github.com/Offensive-AI-Lab/prism
cd prism
uv sync --extra demo
uv run python scripts/download_baselines.py --only ao
uv run python demo/app.py
License
The adapter files are licensed under Apache-2.0. They do not include Qwen3.5-9B weights. The adapted Activation Oracles implementation is MIT, Copyright (c) 2025 Adam Karvonen, and is attributed in the PRISM repository.
References
- Adam Karvonen et al. Activation Oracles: Training and Evaluating LLMs as General-Purpose Activation Explainers.
- Gilad Gressel et al. PRISM: Recovering Instruction Sets from Language Model Activations. EMNLP 2026.
- Library
- peft
- License
- apache-2.0
- Languages
- en
- Created
- Sep 2, 2026
- Updated
- Sep 8, 2026