Modelpublic

Jeesup/svd-safety-l3_swift_remove20_swapgapiter_b010

sourceHugging Facellama3updated 3d agoView on Hugging Face

0likes487downloads

Model Card

svd-safety-l3swiftremove20swapgapiterb010

A Llama-3-8B-Instruct checkpoint compressed with Swift-SVD with LoRA recovery (base checkpoint Jeesup/svd-safety-l3_swift_remove20) to 80.0% of dense parameters, then edited by 10 of 10 rounds of iterative parameter-neutral swap selected by the `gap_iter` rule (up to 0.1% of dense parameters per round; the full run's budget is 1.0%).

This is a research artifact from a study of how SVD compression damages safety behaviour and which component-selection rule best repairs it. It is one cell of a grid over selection rules and budgets; it is not a general-purpose chat model.

Provenance

field	value
base (uncompressed)	`meta-llama/Meta-Llama-3-8B-Instruct`
compression	Swift-SVD with LoRA recovery (base checkpoint `Jeesup/svd-safety-l3_swift_remove20`), 19.96% of parameters removed
selection rule	`gap_iter`
restore budget	1.000% of dense parameters
components restored	9279
components swapped out	9279
resulting parameter fraction	0.8004
seed	42
iterative rounds applied	10 of 10
per-round chunk	0.100% of dense parameters
parameters swapped in	69,743,616 (1.00% of dense projection parameters)
eviction	matrix-local (each insertion evicts a component of the same matrix)
swap value	`insert` (insertion value only; sigma-ordered eviction)

Measured

metric	value
AdvBench ASR (HarmBench judge)	0.0519
StrongREJECT ASR (HarmBench judge)	0.0703
Macro over-refusal (WildGuard)	0.2401
WikiText-2 perplexity	15.3940

Intended use and limitations

This checkpoint exists to measure safety/utility trade-offs under compression. Several arms in the grid are deliberately safety-degraded relative to Llama-3-8B-Instruct: compression alone raises attack-success rate, and the point of the study is to quantify that and test recovery. Treat any given cell as an experimental subject, not as a deployable assistant, and evaluate it yourself before drawing conclusions from it.

Licence

Meta Llama 3 Community License. LICENSE and USE_POLICY.md are included in this repository, and use of this derivative is bound by them. Built with Meta Llama 3.