Paper Citation Record · LEDGER

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates

As of 8 August 2026, this Paper Citation Record lists 35 of 35 outbound references and 0 inbound Pith citation observations for arXiv:2607.28959.

A citation records a reference. It does not transfer a finding from one paper to another.

pith.paper-citation-record.v1
2607.28959 v1

Coverage vector

measured 35 of 35 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-03T16:32:06.668110Z

measured 35 of 35 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-08T06:32:00.761636+00:00

measured 0 of 0 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links

measured 0 of 1 external citation measurements

A source-named dated measurement, never combined with another source.

Source: cited_works

Reference resolution

35 of 35 outbound references displayed

verified exact0
verified fuzzy0
unresolved35
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 2551cf87-a5b1-4484-a124-3b114d97f47c · outbound

This paper cites B Additional Results B.1 More Results Table 5: Additional cross-dataset robustness results on Pythia-1.4B.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates B Additional Results B.1 More Results Table 5: Additional cross-dataset robustness results on Pythia-1.4B

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:06.533912Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3296c2b7-700a-42d3-b72a-66524f9d805d · outbound

This paper cites Transformer feed-forward layers are key-value memories.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Transformer feed-forward layers are key-value memories

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:03.762556Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ab517157-5720-4a3c-b2b4-cab81dd2413f · outbound

This paper cites Have Faith in Faithfulness: Going Beyond Circuit Overlap When Finding Model Mechanisms.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Have Faith in Faithfulness: Going Beyond Circuit Overlap When Finding Model Mechanisms

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:03.929481Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3dcae857-56cf-4d3b-b6a1-080cb97ec7d4 · outbound

This paper cites Impact of positional encoding: Clean and adversarial rademacher complexity for transformers under in-context regression.arXiv preprint arXiv:2512.09275,.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Impact of positional encoding: Clean and adversarial rademacher complexity for transformers under in-context regression.arXiv preprint arXiv:2512.09275,

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:04.041299Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 09b5869f-2af2-4ec9-b95d-8a3193b6c1b5 · outbound

This paper cites Scaling Trends in Language Model Robustness.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Scaling Trends in Language Model Robustness

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:04.086998Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 848ec80b-7426-4632-bcef-803bd1d40a29 · outbound

This paper cites Scaling Laws for Neural Language Models.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Scaling Laws for Neural Language Models

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:04.155055Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 365eb461-c285-4511-9f18-379cb93a83d8 · outbound

This paper cites A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:04.215688Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 613df603-5f04-4fa7-96fe-b57127c83eb5 · outbound

This paper cites Towards understanding jailbreak attacks in llms: A representation space analysis.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Towards understanding jailbreak attacks in llms: A representation space analysis

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:04.296886Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2f6c4c26-d4e2-410a-a026-854060ed12bf · outbound

This paper cites Alignment-constrained dynamic pruning for llms: Identifying and preserving alignment-critical circuits.arXiv preprint arXiv:2511.07482,.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Alignment-constrained dynamic pruning for llms: Identifying and preserving alignment-critical circuits.arXiv preprint arXiv:2511.07482,

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.190586Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 86cc0165-c5b4-41c8-9ded-d4cba665f6ca · outbound

This paper cites Attention sinks and compression valleys in llms are two sides of the same coin.arXiv preprint arXiv:2510.06477,.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Attention sinks and compression valleys in llms are two sides of the same coin.arXiv preprint arXiv:2510.06477,

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.263234Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8db00739-e0ec-4a77-90a9-3e8c425eda61 · outbound

This paper cites A general framework to enhance fine-tuning-based llm unlearning.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates A general framework to enhance fine-tuning-based llm unlearning

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.352575Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d02be9e3-24d2-422e-8a95-88d5772d846a · outbound

This paper cites Open Problems in Mechanistic Interpretability.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Open Problems in Mechanistic Interpretability

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.426090Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a8291210-311e-4da8-86bc-6484b77d4e1e · outbound

This paper cites Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs

Reference 23

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.505938Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fb6ffc55-2fdb-4cc1-b425-57bc77fd86ea · outbound

This paper cites Layer by Layer: Uncovering Hidden Representations in Language Models.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Layer by Layer: Uncovering Hidden Representations in Language Models

Reference 24

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.577174Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2b339f4c-fdf9-465f-bbfb-7c52b605216c · outbound

This paper cites Tensor Trust: Interpretable Prompt Injection Attacks from an Online Game.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Tensor Trust: Interpretable Prompt Injection Attacks from an Online Game

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.653882Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4c8dac68-8c8c-4713-8032-17643f6529db · outbound

This paper cites Steering Language Models With Activation Engineering.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Steering Language Models With Activation Engineering

Reference 26

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.729882Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b46aaab2-d5e1-44f5-83a5-5aed2fab45e8 · outbound

This paper cites Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small

Reference 27

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.802460Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d34c49ad-2ba7-4c9d-b8f7-0132ba070559 · outbound

This paper cites Fast is better than free: Revisiting adversarial training.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Fast is better than free: Revisiting adversarial training

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.877768Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 35456b1c-74db-4559-a505-52e9e6749cf2 · outbound

This paper cites Qwen2.5 Technical Report.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Qwen2.5 Technical Report

Reference 29

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.969389Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cf703032-cc22-4496-aa42-6183a8021c9c · outbound

This paper cites GPTFUZZER: Red Teaming Large Language Models with Auto-Generated Jailbreak Prompts.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates GPTFUZZER: Red Teaming Large Language Models with Auto-Generated Jailbreak Prompts

Reference 30

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:06.059462Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ad144c06-cf9a-4a3f-bdb6-68b24ef6127c · outbound

This paper cites Adversarial Training: A Survey.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Adversarial Training: A Survey

Reference 31

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:06.164646Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7b1d80d9-b084-414a-9a8f-51ecc7ee23d1 · outbound

This paper cites On Prompt-Driven Safeguarding for Large Language Models.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates On Prompt-Driven Safeguarding for Large Language Models

Reference 32

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:06.278220Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation dc43658d-ade5-4f40-ac50-deb247ede3c3 · outbound

This paper cites Representation Engineering: A Top-Down Approach to AI Transparency.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Representation Engineering: A Top-Down Approach to AI Transparency

Reference 33

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:06.408696Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0e9accf4-b6e4-42d9-80ee-6f69908a6df3 · outbound

This paper cites Hence ∥(I−P S)rt∥ ≤1 m0 Mt.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Hence ∥(I−P S)rt∥ ≤1 m0 Mt

Reference 35

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:06.668110Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7f9d2b97-93d2-4982-bbd5-09161d5f1032 · outbound

This paper cites Pruning Convolutional Neural Networks for Resource Efficient Inference.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Pruning Convolutional Neural Networks for Resource Efficient Inference

Reference 2006

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:04.848482Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 55167202-7ded-402f-a8cb-7e661079240d · outbound

This paper cites Towards Deep Learning Models Resistant to Adversarial Attacks.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Towards Deep Learning Models Resistant to Adversarial Attacks

Reference 2011

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:04.443242Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 43a7b00d-903e-46e1-96cd-43356c8c70fb · outbound

This paper cites The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets

Reference 2017

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:04.612787Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 64ff00e6-bf42-4af4-a79f-c8ca2466c3d6 · outbound

This paper cites Softmax is 1/2-lipschitz: A tight bound across all ℓp norms.arXiv preprint arXiv:2510.23012,.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Softmax is 1/2-lipschitz: A tight bound across all ℓp norms.arXiv preprint arXiv:2510.23012,

Reference 2019

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:05.019151Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1eed1946-ed36-448a-8bc9-922f65ba0adf · outbound

This paper cites Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks

Reference 2020

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:03.229149Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9bc1f7c2-9e55-40b0-adbf-cf44297a0133 · outbound

This paper cites The Llama 3 Herd of Models.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates The Llama 3 Herd of Models

Reference 2021

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:03.820766Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 47f5e25e-dfc7-4f77-a259-9fca08aa7157 · outbound

This paper cites Toy Models of Superposition.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Toy Models of Superposition

Reference 2022

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:03.699008Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 59d6cbe8-68ab-46d9-8c3e-f35e0b6373a3 · outbound

This paper cites HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal

Reference 2023

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:04.732124Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a624ce2a-ef17-492b-88e7-954ac420c852 · outbound

This paper cites Mixat: Combining continuous and discrete adversarial training for llms.arXiv preprint arXiv:2505.16947,.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Mixat: Combining continuous and discrete adversarial training for llms.arXiv preprint arXiv:2505.16947,

Reference 2024

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:03.608483Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8ea2f2d8-3de5-4382-8e84-a4fd76f9be3c · outbound

This paper cites Towards understanding safety alignment: A mechanistic perspective from safety neurons.arXiv preprint arXiv:2406.14144,.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Towards understanding safety alignment: A mechanistic perspective from safety neurons.arXiv preprint arXiv:2406.14144,

Reference 2025

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:03.423049Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9c23aac0-fd84-4bc8-8368-9b8bd121bffc · outbound

This paper cites Defending Against Unforeseen Failure Modes with Latent Adversarial Training.

Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogates Defending Against Unforeseen Failure Modes with Latent Adversarial Training

Reference 2026

Resolution: unresolved; no resolver link, observed 2026-08-03T16:32:03.312558Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

No inbound Pith citation observations are available.