Paper Citation Record · LEDGER

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators

As of 19 August 2026, this Paper Citation Record lists 17 of 17 outbound references and 3 inbound Pith citation observations for arXiv:2509.03647.

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

pith.paper-citation-record.v1
2509.03647 v2

Coverage vector

measured 17 of 17 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-05T10:51:44.513163Z

measured 20 of 20 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-18T06:34:40.430872+00:00

measured 3 of 3 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-04T05:36:43.356416Z

measured 0 of 1 external citation measurements

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

Source: arxiv_reference, observed 2026-05-11T00:50:50.435265Z

Reference resolution

17 of 17 outbound references displayed

verified exact0
verified fuzzy0
unresolved17
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 884fc43c-5944-4f2e-a110-dc1f6eefbd75 · outbound

This paper cites Phi-4 Technical Report.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Phi-4 Technical Report

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.470003Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 796afacc-1d83-4d1f-8b57-db5f3b926950 · outbound

This paper cites DeepSeek-V3 Technical Report.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators DeepSeek-V3 Technical Report

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.481798Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 377b2df0-811b-4180-8ce9-86e3676dd4e3 · outbound

This paper cites One-shot Optimized Steering Vectors Mediate Safety-relevant Behaviors in LLMs.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators One-shot Optimized Steering Vectors Mediate Safety-relevant Behaviors in LLMs

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.484491Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 61946c37-67fa-43da-8bd9-ec4f82121088 · outbound

This paper cites The Llama 3 Herd of Models.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators The Llama 3 Herd of Models

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.487461Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 96025470-ecf9-4e35-a854-5c51dd027bba · outbound

This paper cites A Survey on LLM-as-a-Judge.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators A Survey on LLM-as-a-Judge

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.490091Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation df1eccca-07c1-48a8-b0c5-70f8f8d8a90d · outbound

This paper cites LLMs as Narcissistic Evaluators: When Ego Inflates Evaluation Scores.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators LLMs as Narcissistic Evaluators: When Ego Inflates Evaluation Scores

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.492755Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d3be7bdb-19ae-4b70-853f-6e0d66ffbc5c · outbound

This paper cites LLM Evaluators Recognize and Favor Their Own Generations.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators LLM Evaluators Recognize and Favor Their Own Generations

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.497879Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2a05c428-d11d-4870-9ba0-98eebba3cf36 · outbound

This paper cites Direct Preference Optimization: Your Language Model is Secretly a Reward Model.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Direct Preference Optimization: Your Language Model is Secretly a Reward Model

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.500620Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6bdeb55c-ce83-427d-969d-5e8c8a117827 · outbound

This paper cites Rerouting LLM Routers.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Rerouting LLM Routers

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.503048Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c33fa0e4-a6f3-491d-9d2b-c883088b9adb · outbound

This paper cites Self-Preference Bias in LLM-as-a-Judge.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Self-Preference Bias in LLM-as-a-Judge

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.505631Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation abb5264d-5f38-4023-999b-72fc205319bd · outbound

This paper cites Justice or Prejudice? Quantifying Biases in LLM-as-a-Judge.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Justice or Prejudice? Quantifying Biases in LLM-as-a-Judge

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.508182Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7bd8e9d5-44b3-46e2-9d24-04caeee55b22 · outbound

This paper cites Leveraging Uncertainty Estimation for Efficient LLM Routing.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Leveraging Uncertainty Estimation for Efficient LLM Routing

Reference 16

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.510706Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 95f663a8-0f6f-434c-a24f-bfc367ab9160 · outbound

This paper cites Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.513163Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 47a93026-2b4c-4a59-b9cc-8bc98faef0b3 · outbound

This paper cites Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarization

Reference 2018

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.495367Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8a537809-b3ad-4d46-a0ed-39b94ad498ba · outbound

This paper cites VisIT-Bench: A Benchmark for Vision-Language Instruction Following Inspired by Real-World Use.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators VisIT-Bench: A Benchmark for Vision-Language Instruction Following Inspired by Real-World Use

Reference 2023

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.475907Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d673015e-3920-4df5-beac-72cb4138ac6a · outbound

This paper cites Do LLM Evaluators Pre- fer Themselves for a Reason?, April 2025a.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Do LLM Evaluators Pre- fer Themselves for a Reason?, April 2025a

Reference 2024

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.478823Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 81a2150a-7c62-44be-a267-e3c553b4f0de · outbound

This paper cites Inspection and Control of Self-Generated-Text Recognition Ability in Llama3-8b-Instruct.

Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators Inspection and Control of Self-Generated-Text Recognition Ability in Llama3-8b-Instruct

Reference 2025

Resolution: unresolved; no resolver link, observed 2026-08-05T10:51:44.473173Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation 7cb81bf0-2e4d-4052-936b-77c570e966e6 · inbound

Self-Preference Bias in Rubric-Based Evaluation of Large Language Models cites this paper.

Self-Preference Bias in Rubric-Based Evaluation of Large Language Models Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators

Reference 13

Resolution: verified exact; arxiv_id, observed 2026-06-24T01:14:21.924280Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 02ac22cc-c851-49b1-8ac1-3267d1deaeb0 · inbound

Self-Preference Bias in Rubric-Based Evaluation of Large Language Models cites this paper.

Self-Preference Bias in Rubric-Based Evaluation of Large Language Models Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-02T16:40:55.442545Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6ab5c2fe-4aef-4e52-8ce4-19efc153bb24 · inbound

Self-Preference Bias in Rubric-Based Evaluation of Large Language Models cites this paper.

Self-Preference Bias in Rubric-Based Evaluation of Large Language Models Breaking the Mirror: Activation-Based Mitigation of Self-Preference in LLM Evaluators

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-04T05:36:43.356416Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.