HFEPX · Eval paper review
Miao Yu, Siyuan Fu, Moayad Aloqaily, Zhenhong Zhou +5 more
Published
Mar 24, 2026
Citations
0
Trust level
Provisional
Usefulness score
Unavailable
Extraction confidence
0% (Provisional)
Derived from abstract and metadata only.
Signals refreshed
Mar 24, 2026
Signal extraction is still processing. This page currently shows metadata-first guidance until structured protocol fields are ready.
This page is a lightweight research summary built from the abstract and metadata while deeper extraction catches up.
All signals on this page are inferred from the abstract only and may be inaccurate. Do not use this page as a primary protocol reference.
Best use
Background context only
Use if you need
A provisional background reference while structured extraction finishes.
What to verify
Read the full paper before copying any benchmark, metric, or protocol choices.
Main weakness
This page is still relying on abstract and metadata signals, not a fuller protocol read.
Eval-fit score is unavailable until extraction completes.
If you are doing eval pipeline work, start here
Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components. However, existing safety attribution methods struggle with generalization and reliability due to their reliance on heuristic, domain-specific metrics and search algorithms. To address this, we propose \ourmethod, a unified safety interpretability framework that identifies functionally complete safety circuits in LLMs via optimization. Unlike methods focusing on isolated heads or neurons, \ourmethod introduces differentiable binary masks to extract multi-granular circuits through gradient descent on safety datasets, while integrates Safety Circuit Tuning to utilize these sparse circuits for efficient safety fine-tuning. We validate \ourmethod in two key scenarios in LLM safety: \textbf{(1) backdoor attacks}, identifying a backdoor circuit with 0.42\% sparsity, whose ablation eradicates the Attack Success Rate (ASR) from 100\% $\to$ 0.4\% while retaining over 99\% general utility; \textbf{(2) safety alignment}, localizing an alignment circuit with 3.03\% heads and 0.79\% neurons, whose removal spikes ASR from 0.8\% $\to$ 96.9\%, whereas excluding this circuit during helpfulness fine-tuning maintains 96.5\% safety retention.
These are the protocol signals we could actually recover from the available paper metadata. Use them to decide whether this paper is worth deeper reading.
None explicit
No explicit feedback protocol extracted.
"Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components."
None explicit
Validate eval design from full paper text.
"Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components."
Not reported
No explicit QC controls found.
"Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components."
Not extracted
No benchmark anchors detected.
"Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components."
Not extracted
No metric anchors detected.
"Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components."
Unknown
Rater source not explicitly reported.
"Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components."
This page is using abstract-level cues only right now. Treat the signals below as provisional.
Evaluation fields are inferred from the abstract only.
Mechanistic interpretability reveals that safety-critical behaviors (e.g., alignment, jailbreak, backdoor) in Large Language Models (LLMs) are grounded in specialized functional components.
Based on abstract + metadata only. Check the source paper before making high-confidence protocol decisions.
