Paper Citation Record · LEDGER

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement

As of 21 August 2026, this Paper Citation Record lists 16 of 16 outbound references and 1 inbound Pith citation observation for arXiv:2507.18742.

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

pith.paper-citation-record.v1
2507.18742 v1

Coverage vector

measured 16 of 16 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-15T18:14:14.618830Z

measured 17 of 17 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-21T06:32:19.484+00:00

measured 1 of 1 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-07-31T18:50:51.463421Z

measured 0 of 1 external citation measurements

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

Source: cited_works

Reference resolution

16 of 16 outbound references displayed

verified exact1
verified fuzzy2
unresolved13
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 92300e3e-88d9-4da2-9150-53f00ceaa254 · outbound

This paper cites Concrete Problems in AI Safety.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Concrete Problems in AI Safety

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.543125Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7a99be13-5085-41e5-b4c9-ca4d169b42a0 · outbound

This paper cites Constitutional AI: Harmlessness from AI Feedback.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Constitutional AI: Harmlessness from AI Feedback

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.548634Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d580fcf6-dbc9-40b4-99f7-9cdf28ae70b5 · outbound

This paper cites Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.553499Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 876c917d-e201-45f0-9bcf-bc7fe7763644 · outbound

This paper cites Meta SC : Test-time safety specification optimization for language models.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Meta SC : Test-time safety specification optimization for language models

Reference 4

Resolution: verified fuzzy; raw_fallback, observed 2026-08-15T18:14:14.844820Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.

Observation cb297aff-7114-465e-9264-f0fe276ab9d5 · outbound

This paper cites Self-refine: Iterative refinement with self-feedback.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Self-refine: Iterative refinement with self-feedback

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.563129Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bd032ca6-6ea0-4d40-b09f-50bf906ddaa2 · outbound

This paper cites Honesty to Subterfuge: In-Context Reinforcement Learning Can Make Honest Models Reward Hack.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Honesty to Subterfuge: In-Context Reinforcement Learning Can Make Honest Models Reward Hack

Reference 6

Resolution: verified exact; local_arxiv, observed 2026-08-15T18:14:14.688788Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.

Observation b2ae65ad-358f-4117-a267-578ab09b979c · outbound

This paper cites Training language models to follow instructions with human feedback.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Training language models to follow instructions with human feedback

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.573132Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e6a82850-c8ad-4539-8030-f72180d66015 · outbound

This paper cites The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.578328Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d6dd4117-70e2-43b0-95cb-a3c55cb0a460 · outbound

This paper cites Feedback loops with language models drive in-context reward hacking.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Feedback loops with language models drive in-context reward hacking

Reference 9

Resolution: verified fuzzy; raw_fallback, observed 2026-08-15T18:14:14.812064Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.

Observation 2461dfed-3890-4700-82f6-afee731e73bf · outbound

This paper cites an unresolved cited work.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Unresolved cited work

Reference 10

Resolution: unresolved; raw_fallback, observed 2026-08-15T18:14:14.796642Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.

Observation 9d525264-2aaa-41aa-911e-33df0121d2ab · outbound

This paper cites A Theoretical Understanding of Self-Correction through In-context Alignment.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement A Theoretical Understanding of Self-Correction through In-context Alignment

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.591491Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7bdeba25-7c1a-4740-a7a3-8f016ad5ce7c · outbound

This paper cites Reward hacking in reinforcement learning.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Reward hacking in reinforcement learning

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.596987Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e84dd880-69c4-4c75-ae14-3285da599b18 · outbound

This paper cites write newline.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement write newline

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.603803Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation be3bb819-8425-4c9f-9981-628eba89f808 · outbound

This paper cites @esa (Ref.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement @esa (Ref

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.609788Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d0adc396-1514-426a-a319-edb4bf246f4f · outbound

This paper cites an unresolved cited work.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Unresolved cited work

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-15T18:14:14.614450Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2e0ce294-199e-42bf-ba85-ded73cc29600 · outbound

This paper cites an unresolved cited work.

Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement Unresolved cited work

Reference 16

Resolution: unresolved; raw_fallback, observed 2026-08-15T18:14:14.745525Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-21T06:32:19.484+00:00.

Pith citing papers

Observation b6fb5169-8fe0-4d29-8e11-b3bf1416498c · inbound

Self-Authored Verification Is Unreliable in Heuristic Self-Improving Agents cites this paper.

Self-Authored Verification Is Unreliable in Heuristic Self-Improving Agents Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement

Reference 5

Resolution: unresolved; no resolver link, observed 2026-07-31T18:50:51.463421Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.