Instructions to use Jeesup/svdsafety_l2_ablate_random64 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Jeesup/svdsafety_l2_ablate_random64 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Jeesup/svdsafety_l2_ablate_random64")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Jeesup/svdsafety_l2_ablate_random64")
model = AutoModelForCausalLM.from_pretrained("Jeesup/svdsafety_l2_ablate_random64", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Jeesup/svdsafety_l2_ablate_random64 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Jeesup/svdsafety_l2_ablate_random64"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Jeesup/svdsafety_l2_ablate_random64",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Jeesup/svdsafety_l2_ablate_random64

SGLang

How to use Jeesup/svdsafety_l2_ablate_random64 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Jeesup/svdsafety_l2_ablate_random64" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Jeesup/svdsafety_l2_ablate_random64",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Jeesup/svdsafety_l2_ablate_random64" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Jeesup/svdsafety_l2_ablate_random64",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Jeesup/svdsafety_l2_ablate_random64 with Docker Model Runner:
```
docker model run hf.co/Jeesup/svdsafety_l2_ablate_random64
```

svdsafety_l2_ablate_random64 / README.md

Jeesup

Delta-subspace ablation checkpoint and evaluation

a9483ff verified about 2 months ago

preview code

Raw History Blame Contribute Delete

2.23 kB

---
license: llama2
base_model: meta-llama/Llama-2-7b-chat-hf
library_name: transformers
pipeline_tag: text-generation
tags: [llama2, safety, interpretability, ablation, svd]
---
# svdsafety_l2_ablate_random64
**A safety-ablation research artifact, not a chat model.** Control for `ablate`. Removes **64 uniformly random** singular components of `dW`, norm-matched to the top-64 block (seed 42).
Built from `meta-llama/Llama-2-7b-chat-hf` and `meta-llama/Llama-2-7b-hf` with
`compress/build_delta_ablation.py`, k=64, all 224 projections edited, seed 42.
The weight perturbation `||W' - W||_F / ||W||_F` is **0.02423** averaged over
matrices.
## Measured behaviour
AdvBench (520 prompts), greedy chat decoding, judged by
`cais/HarmBench-Llama-2-13b-cls`:
| | this checkpoint | dense Llama-2-7b-chat |
|---|---:|---:|
| attack success rate | **0.0058** | 0.0019 |
| refusal rate (harmful) | **0.992** | 0.994 |
| degenerate output rate | 0.004 | 0.000 |
## Why these exist
Compressing a chat model by singular value discards its alignment update almost
by construction: `sigma(dW)` is orders of magnitude below `sigma(W_base)`, and
the leading 64 directions of `dW` hold only 15% of its energy. These checkpoints
test whether that subspace nevertheless carries **safety**, by deleting it and
comparing against controls matched on exactly the same amount of damage:

| edit (k=64; perturbation 0.02423 for all three) | ASR | refusal |
|---|---:|---:|
| remove **top**-64 of dW | 0.0962 | 0.831 |
| remove **bottom**-64 of dW | 0.0038 | 0.994 |
| remove **random** 64 of dW | 0.0058 | 0.992 |

Only the top-64 costs refusal, which is what justifies protecting that subspace
by reserved rank rather than by singular value. The subspaces themselves are at
[`Jeesup/svdsafety_l2_delta_basis`](https://huggingface.co/Jeesup/svdsafety_l2_delta_basis).

## Intended use

Measuring how alignment survives low-rank compression. Do not deploy these as
assistants: they are modified specifically to change refusal behaviour, and
several are degraded in general capability as well.

## Licence

Llama 2 Community License. `LICENSE.txt` and `USE_POLICY.md` are included here
and use of this derivative is bound by both. Built with Llama 2.

	---
	license: llama2
	base_model: meta-llama/Llama-2-7b-chat-hf
	library_name: transformers
	pipeline_tag: text-generation
	tags: [llama2, safety, interpretability, ablation, svd]
	---

	# svdsafety_l2_ablate_random64

	A safety-ablation research artifact, not a chat model. Control for `ablate`. Removes 64 uniformly random singular components of `dW`, norm-matched to the top-64 block (seed 42).

	Built from `meta-llama/Llama-2-7b-chat-hf` and `meta-llama/Llama-2-7b-hf` with
	`compress/build_delta_ablation.py`, k=64, all 224 projections edited, seed 42.
	The weight perturbation `\|\|W' - W\|\|_F / \|\|W\|\|_F` is 0.02423 averaged over
	matrices.

	## Measured behaviour

	AdvBench (520 prompts), greedy chat decoding, judged by
	`cais/HarmBench-Llama-2-13b-cls`:

	\| \| this checkpoint \| dense Llama-2-7b-chat \|
	\|---\|---:\|---:\|
	\| attack success rate \| 0.0058 \| 0.0019 \|
	\| refusal rate (harmful) \| 0.992 \| 0.994 \|
	\| degenerate output rate \| 0.004 \| 0.000 \|



	## Why these exist

	Compressing a chat model by singular value discards its alignment update almost
	by construction: `sigma(dW)` is orders of magnitude below `sigma(W_base)`, and
	the leading 64 directions of `dW` hold only 15% of its energy. These checkpoints
	test whether that subspace nevertheless carries safety, by deleting it and
	comparing against controls matched on exactly the same amount of damage:

	\| edit (k=64; perturbation 0.02423 for all three) \| ASR \| refusal \|
	\|---\|---:\|---:\|
	\| remove top-64 of dW \| 0.0962 \| 0.831 \|
	\| remove bottom-64 of dW \| 0.0038 \| 0.994 \|
	\| remove random 64 of dW \| 0.0058 \| 0.992 \|

	Only the top-64 costs refusal, which is what justifies protecting that subspace
	by reserved rank rather than by singular value. The subspaces themselves are at
	[`Jeesup/svdsafety_l2_delta_basis`](https://huggingface.co/Jeesup/svdsafety_l2_delta_basis).

	## Intended use

	Measuring how alignment survives low-rank compression. Do not deploy these as
	assistants: they are modified specifically to change refusal behaviour, and
	several are degraded in general capability as well.

	## Licence

	Llama 2 Community License. `LICENSE.txt` and `USE_POLICY.md` are included here
	and use of this derivative is bound by both. Built with Llama 2.