Paper Citation Record · LEDGER

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples

As of 9 August 2026, this Paper Citation Record lists 94 of 94 outbound references and 5 inbound Pith citation observations for arXiv:2502.09650.

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

pith.paper-citation-record.v1
2502.09650 v2

Coverage vector

measured 94 of 94 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-08T11:57:12.088561Z

measured 99 of 99 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 5 of 5 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-06T18:55:17.631077Z

measured 0 of 1 external citation measurements

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

Source: arxiv_reference, observed 2026-07-03T00:37:30.066482Z

Reference resolution

94 of 94 outbound references displayed

verified exact1
verified fuzzy35
unresolved58
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 0a933d5a-ad80-4ede-bab5-ed53bf83892e · outbound

This paper cites write newline.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples write newline

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.661466Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 655266a0-cd06-4946-add8-dca8dff3a089 · outbound

This paper cites Llama 3 model card.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Llama 3 model card

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.667419Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 57bbada3-4c13-49f8-afac-91e6e5a6ed21 · outbound

This paper cites M., Longpre, S., Lambert, N., Wang, X., Muennighoff, N., Hou, B., Pan, L., Jeong, H., Raffel, C., Chang, S., Hashimoto, T., and Wang, W.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples M., Longpre, S., Lambert, N., Wang, X., Muennighoff, N., Hou, B., Pan, L., Jeong, H., Raffel, C., Chang, S., Hashimoto, T., and Wang, W

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.671948Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6e6e59b9-9d2e-404c-ac45-9fdc6889f447 · outbound

This paper cites Perplexed by Perplexity: Perplexity-Based Data Pruning With Small Reference Models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Perplexed by Perplexity: Perplexity-Based Data Pruning With Small Reference Models

Reference 4

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.676305Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 605a9ffb-e7c1-4037-9535-20a6c4ab1b3f · outbound

This paper cites Ultrafeedback binarized preferences-cleaned.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Ultrafeedback binarized preferences-cleaned

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.681101Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4dcbc8d6-dd6e-4c34-8d7a-76acbff3d966 · outbound

This paper cites A General Language Assistant as a Laboratory for Alignment.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples A General Language Assistant as a Laboratory for Alignment

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.685443Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f0859dfe-08ef-4b56-ad86-17e0e915f779 · outbound

This paper cites Teaching by examples: Implications for the process of category acquisition.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Teaching by examples: Implications for the process of category acquisition

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.690190Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9da6ea09-c173-49e0-8e8d-8d975e47b590 · outbound

This paper cites G., Guo, Z.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples G., Guo, Z

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.694676Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b71b82f7-5985-4a1c-afcd-07679dda7d93 · outbound

This paper cites Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.698975Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e261bedc-56bf-4bcc-9069-07ce8a9f60ab · outbound

This paper cites an unresolved cited work.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Unresolved cited work

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.706330Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 32f02130-c072-4f41-ad63-87d146a7684f · outbound

This paper cites Open LLM leaderboard.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Open LLM leaderboard

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.710699Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 07777237-1d3c-4473-aa2d-ea91f2173d2d · outbound

This paper cites Curriculum learning.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Curriculum learning

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.714968Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cc0cf199-6327-40dd-b7f4-a3b1c1db935c · outbound

This paper cites D., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples D., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.719298Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5bb97408-e321-487e-9c3b-d58ac09da152 · outbound

This paper cites Curriculum learning for language modeling.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Curriculum learning for language modeling

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.723302Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4fe5805a-6445-47aa-b221-09d44d6ab5f2 · outbound

This paper cites Instruction Mining: Instruction Data Selection for Tuning Large Language Models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Instruction Mining: Instruction Data Selection for Tuning Large Language Models

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.728183Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 62fe3a44-be84-43aa-a070-2c43626dc536 · outbound

This paper cites Does the order of training samples matter? improving neural data-to-text generation with curriculum learning.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Does the order of training samples matter? improving neural data-to-text generation with curriculum learning

Reference 16

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.733077Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ee77afaa-5587-488d-9163-7eb5f02abd9e · outbound

This paper cites Alpagasus: Training a better alpaca with fewer data.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Alpagasus: Training a better alpaca with fewer data

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.737460Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 031b1471-498f-4f13-afa7-740b6b25be23 · outbound

This paper cites Skill-it! a data-driven skills framework for understanding and training language models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Skill-it! a data-driven skills framework for understanding and training language models

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.742169Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5a9e2a0a-dfd6-4f37-ba10-92c534bb9e07 · outbound

This paper cites Sdformer: Similarity-driven discrete transformer for time series generation.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Sdformer: Similarity-driven discrete transformer for time series generation

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.746264Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 73615786-de88-418e-a2d8-f9e4bfe2ee1f · outbound

This paper cites R., Kini, A., and Natarajan, N.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples R., Kini, A., and Natarajan, N

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.309048Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation d01213d0-5a6f-4594-a656-f9104ba1c249 · outbound

This paper cites F., Leike, J., Brown, T., Martic, M., Legg, S., and Amodei, D.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples F., Leike, J., Brown, T., Martic, M., Legg, S., and Amodei, D

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.759750Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a5c8f22b-a6b8-4bf2-afed-ea171b7136f6 · outbound

This paper cites Training Verifiers to Solve Math Word Problems.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Training Verifiers to Solve Math Word Problems

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.764207Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 04942a48-e175-4873-892f-abc86e3a6375 · outbound

This paper cites Reverse Stable Diffusion: What prompt was used to generate this image?.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Reverse Stable Diffusion: What prompt was used to generate this image?

Reference 23

Resolution: verified exact; local_arxiv, observed 2026-08-08T11:57:12.470267Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation dd1ed5de-a3ed-4067-a9a1-abb6b5d77eed · outbound

This paper cites Curriculum Direct Preference Optimization for Diffusion and Consistency Models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Curriculum Direct Preference Optimization for Diffusion and Consistency Models

Reference 24

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.775726Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2ef54133-80ae-4ec1-a897-b8bae6146a04 · outbound

This paper cites MoDS: Model-oriented Data Selection for Instruction Tuning.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples MoDS: Model-oriented Data Selection for Instruction Tuning

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.781005Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 995355fc-80ec-40fa-8905-d823662788d0 · outbound

This paper cites The Llama 3 Herd of Models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples The Llama 3 Herd of Models

Reference 26

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.785598Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 175afe08-8912-4b17-96bd-099c513ca4ee · outbound

This paper cites Length-controlled alpacaeval: A simple debiasing of automatic evaluators.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Length-controlled alpacaeval: A simple debiasing of automatic evaluators

Reference 27

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.288253Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 75bfca09-7ec5-4c8a-9aeb-da243389824f · outbound

This paper cites Dsdm: Model-aware dataset selection with datamodels.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Dsdm: Model-aware dataset selection with datamodels

Reference 28

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.274675Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 5c6ec3db-1c06-45e6-a387-f3711f9458b9 · outbound

This paper cites Understanding dataset difficulty with \ V \ -usable information.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Understanding dataset difficulty with \ V \ -usable information

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.261627Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 2eebdb39-6605-40ea-b02c-2819b49dbc07 · outbound

This paper cites Model alignment as prospect theoretic optimization.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Model alignment as prospect theoretic optimization

Reference 30

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.248336Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation a14c0fb6-3e55-43db-afe2-d1243e18a279 · outbound

This paper cites Impact of preference noise on the alignment performance of generative language models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Impact of preference noise on the alignment performance of generative language models

Reference 31

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.234931Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation f3e08ea9-aae9-4812-9fae-d47368c74181 · outbound

This paper cites Understanding finetuning for factual knowledge extraction.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Understanding finetuning for factual knowledge extraction

Reference 32

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.222187Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 00099db9-8739-4b1c-813b-d2136ec4efc6 · outbound

This paper cites ORPO: Monolithic Preference Optimization without Reference Model.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples ORPO: Monolithic Preference Optimization without Reference Model

Reference 33

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.814773Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1ddbfca8-8e12-4ce2-ab49-3c98d34eabd9 · outbound

This paper cites J., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., Chen, W., et al.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples J., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., Chen, W., et al

Reference 34

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.819203Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6793c83e-ec25-47ae-86ab-202543295f95 · outbound

This paper cites R., and Kwon, D.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples R., and Kwon, D

Reference 35

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.200527Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 95d7a382-e5e9-4146-b24d-fda38f6883ba · outbound

This paper cites Mistral 7B.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Mistral 7B

Reference 36

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.826904Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 922a2571-2e4d-46d1-8fa7-ff119a734921 · outbound

This paper cites an unresolved cited work.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Unresolved cited work

Reference 37

Resolution: unresolved; raw_fallback, observed 2026-08-08T11:57:13.187780Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation c2e50a66-9019-4125-8d4a-f49120d02f91 · outbound

This paper cites FastText.zip: Compressing text classification models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples FastText.zip: Compressing text classification models

Reference 38

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.834633Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 22f798a6-2699-4fbd-a2c0-01765da5c2ad · outbound

This paper cites Strategic Data Ordering: Enhancing Large Language Model Performance through Curriculum Learning.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Strategic Data Ordering: Enhancing Large Language Model Performance through Curriculum Learning

Reference 39

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.838854Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 407c4005-7e06-40a9-ad05-bd90e5f34dfc · outbound

This paper cites an unresolved cited work.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Unresolved cited work

Reference 40

Resolution: unresolved; raw_fallback, observed 2026-08-08T11:57:13.174778Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 6c5f8b8d-a28b-42e0-a1f5-5f11f87b7054 · outbound

This paper cites The bigscience roots corpus: A 1.6 tb composite multilingual dataset.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples The bigscience roots corpus: A 1.6 tb composite multilingual dataset

Reference 41

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.160470Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation a73ca80c-d3fb-4b66-b229-20d104962919 · outbound

This paper cites Deduplicating Training Data Makes Language Models Better.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Deduplicating Training Data Makes Language Models Better

Reference 42

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.850578Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3fd1f474-239e-468d-8eca-f904ba2e90a2 · outbound

This paper cites From quantity to quality: Boosting llm performance with self-guided data selection for instruction tuning.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples From quantity to quality: Boosting llm performance with self-guided data selection for instruction tuning

Reference 43

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.854392Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6bdd99c2-b198-4483-a296-4380124dc056 · outbound

This paper cites From Crowdsourced Data to High-Quality Benchmarks: Arena-Hard and BenchBuilder Pipeline.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples From Crowdsourced Data to High-Quality Benchmarks: Arena-Hard and BenchBuilder Pipeline

Reference 44

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.858507Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f3ac5268-774e-4fca-8653-f613abf7228d · outbound

This paper cites Not all tokens are what you need for pretraining.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Not all tokens are what you need for pretraining

Reference 45

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.862678Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9c87cbab-1c7c-45b6-82ff-9fc427a1c648 · outbound

This paper cites High dimensional robust m-estimation: Arbitrary corruption and heavy tails.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples High dimensional robust m-estimation: Arbitrary corruption and heavy tails

Reference 46

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.128916Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation a70c0d2d-c428-4ac4-987a-b2fe3b1531d2 · outbound

This paper cites What makes good data for alignment? a comprehensive study of automatic data selection in instruction tuning.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples What makes good data for alignment? a comprehensive study of automatic data selection in instruction tuning

Reference 47

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.114562Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 08b2976a-b413-4734-9fab-b0c9f6b01ff6 · outbound

This paper cites Simpo: Simple preference optimization with a reference-free reward.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Simpo: Simple preference optimization with a reference-free reward

Reference 48

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.100378Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation c85bd4a2-e299-4a9c-8397-10f4a19d2911 · outbound

This paper cites and Sachdeva, B.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples and Sachdeva, B

Reference 49

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.086970Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation e7d6aa9d-decf-4b25-9720-4277ed638ab2 · outbound

This paper cites A note on dpo with noisy preferences and relationship to ipo, 2023.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples A note on dpo with noisy preferences and relationship to ipo, 2023

Reference 50

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.073007Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 7b6659a8-b42c-4ddf-9082-713bce8805ee · outbound

This paper cites Orca: Progressive Learning from Complex Explanation Traces of GPT-4.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Orca: Progressive Learning from Complex Explanation Traces of GPT-4

Reference 51

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.891521Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 392909e0-5bac-496e-aa51-7cebca0482c2 · outbound

This paper cites Length-based curriculum learning for efficient pre-training of language models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Length-based curriculum learning for efficient pre-training of language models

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.058550Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation ec70e3df-60ad-476f-b01a-f56379d30e89 · outbound

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

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Training language models to follow instructions with human feedback

Reference 53

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.900352Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 69a027fa-09de-4856-a636-9c3d356ad18b · outbound

This paper cites Disentangling length from quality in direct preference optimization.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Disentangling length from quality in direct preference optimization

Reference 54

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.035108Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 011012fb-a26c-4fd3-b3be-7af78c817c61 · outbound

This paper cites Curry-DPO: Enhancing Alignment using Curriculum Learning & Ranked Preferences.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Curry-DPO: Enhancing Alignment using Curriculum Learning & Ranked Preferences

Reference 55

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.908968Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9182f382-07a8-48d3-8ec8-8b6dee8cdf33 · outbound

This paper cites The refinedweb dataset for falcon llm: Outperforming curated corpora with web data only.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples The refinedweb dataset for falcon llm: Outperforming curated corpora with web data only

Reference 56

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.020270Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation ae81375a-4ac7-491b-aacf-0eac71313965 · outbound

This paper cites Competence-based Curriculum Learning for Neural Machine Translation.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Competence-based Curriculum Learning for Neural Machine Translation

Reference 57

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.918643Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9a191d7d-59e8-4d31-80bb-c97fca1aaa28 · outbound

This paper cites Unleashing the power of data tsunami: A comprehensive survey on data assessment and selection for instruction tuning of language models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Unleashing the power of data tsunami: A comprehensive survey on data assessment and selection for instruction tuning of language models

Reference 58

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:13.005599Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 083b9713-f2ce-498c-8666-2674a89f69b8 · outbound

This paper cites D., Ermon, S., and Finn, C.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples D., Ermon, S., and Finn, C

Reference 59

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.931086Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 08e846eb-3a94-4287-ad2b-72e4dabb1f8e · outbound

This paper cites Does the Definition of Difficulty Matter? Scoring Functions and their Role for Curriculum Learning.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Does the Definition of Difficulty Matter? Scoring Functions and their Role for Curriculum Learning

Reference 60

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.934793Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a7960c65-ab17-4a1c-9859-e573a68aaf78 · outbound

This paper cites Unintentional unalignment: Likelihood displacement in direct preference optimization.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Unintentional unalignment: Likelihood displacement in direct preference optimization

Reference 61

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.978007Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation d11b659b-cd51-414d-a25b-32c931f16615 · outbound

This paper cites and Xing, E.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples and Xing, E

Reference 62

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.962510Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation a8a25d3b-ed1a-4554-b97f-be09715256ae · outbound

This paper cites and Sanghavi, S.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples and Sanghavi, S

Reference 63

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.949944Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 979a4161-c013-4ee9-ac56-681e8739204d · outbound

This paper cites less is more.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples less is more

Reference 64

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.936911Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 5531633e-afad-4aa1-bec4-d018de30fe39 · outbound

This paper cites Dataset Cartography: Mapping and Diagnosing Datasets with Training Dynamics.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Dataset Cartography: Mapping and Diagnosing Datasets with Training Dynamics

Reference 65

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.955008Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ae8a458b-9b4b-4162-8f9a-37e872afea67 · outbound

This paper cites Txt360: A top-quality llm pre-training dataset requires the perfect blend, 2024.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Txt360: A top-quality llm pre-training dataset requires the perfect blend, 2024

Reference 66

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.923219Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation f490e476-e0d8-4040-9af3-5d3f01dde2e3 · outbound

This paper cites T., Fu, J., Phan, M.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples T., Fu, J., Phan, M

Reference 67

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.910127Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 80644951-0cad-452e-a7d7-e00693aa1dc1 · outbound

This paper cites Gemma 2: Improving Open Language Models at a Practical Size.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Gemma 2: Improving Open Language Models at a Practical Size

Reference 68

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.967745Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cb5c15be-f400-401a-b886-f8b25640dbec · outbound

This paper cites P., and Ferrari, V.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples P., and Ferrari, V

Reference 69

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.973198Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 870ea2fd-4993-429b-b46a-daf4079775d1 · outbound

This paper cites Zephyr: Direct Distillation of LM Alignment.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Zephyr: Direct Distillation of LM Alignment

Reference 70

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.978495Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b94a2143-d519-4a37-9fa7-fbb3ac0946b1 · outbound

This paper cites Secrets of RLHF in Large Language Models Part II: Reward Modeling.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Secrets of RLHF in Large Language Models Part II: Reward Modeling

Reference 71

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.983807Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 90a7941f-47b3-48c6-a163-0442e9a72dc6 · outbound

This paper cites Step-on-feet tuning: Scaling self-alignment of LLM s via bootstrapping.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Step-on-feet tuning: Scaling self-alignment of LLM s via bootstrapping

Reference 72

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.888359Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 2bf987ba-2635-4a5b-b5df-1583e60a6e02 · outbound

This paper cites Interpretable preferences via multi-objective reward modeling and mixture-of-experts.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Interpretable preferences via multi-objective reward modeling and mixture-of-experts

Reference 73

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.875678Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 49a71b7f-cfeb-401b-9744-2c979fcd276a · outbound

This paper cites A., Beltagy, I., et al.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples A., Beltagy, I., et al

Reference 74

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:11.997368Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 940ba6b1-8a36-4822-86f9-46e1db1009d1 · outbound

This paper cites Ethical and social risks of harm from Language Models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Ethical and social risks of harm from Language Models

Reference 75

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.001804Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0b6dea10-5fa3-4ed9-8f74-42e7b1419e38 · outbound

This paper cites CCNet: Extracting High Quality Monolingual Datasets from Web Crawl Data.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples CCNet: Extracting High Quality Monolingual Datasets from Web Crawl Data

Reference 76

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.006202Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f6c989b5-6636-4f75-a4db-1dd8a05d1b25 · outbound

This paper cites QuRating : Selecting high-quality data for training language models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples QuRating : Selecting high-quality data for training language models

Reference 77

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.853177Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation ca3dbf38-26b2-4c2f-8cf0-69da62b1bb1e · outbound

This paper cites Curriculum Learning with Quality-Driven Data Selection.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Curriculum Learning with Quality-Driven Data Selection

Reference 78

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.015145Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f622c6a1-b7c3-4f5a-8aca-07d4679662ef · outbound

This paper cites Self-Evolved Diverse Data Sampling for Efficient Instruction Tuning.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Self-Evolved Diverse Data Sampling for Efficient Instruction Tuning

Reference 79

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.021189Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8f2607de-027a-4b4c-b720-0a72b82dd6ed · outbound

This paper cites When do curricula work? In ICLR, 2021.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples When do curricula work? In ICLR, 2021

Reference 80

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.839294Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 6df0da23-b153-4c5a-8ac2-fa80700025b4 · outbound

This paper cites Rethinking Data Selection at Scale: Random Selection is Almost All You Need.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Rethinking Data Selection at Scale: Random Selection is Almost All You Need

Reference 81

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.030030Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7e51ab70-6814-4151-b81d-0f1e62d84023 · outbound

This paper cites an unresolved cited work.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Unresolved cited work

Reference 82

Resolution: unresolved; raw_fallback, observed 2026-08-08T11:57:12.825816Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation b95b6ad0-937c-46af-9755-55ed9ae87e2f · outbound

This paper cites Rethinking the instruction quality: Lift is what you need.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Rethinking the instruction quality: Lift is what you need

Reference 83

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.812254Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation eadf9e32-3fdb-4a20-a638-ca514234b2e6 · outbound

This paper cites Qwen2.5 Technical Report.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Qwen2.5 Technical Report

Reference 84

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.046156Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7f9bfb16-3836-4a50-9d59-965f0af0bbab · outbound

This paper cites Mates: Model-aware data selection for efficient pretraining with data influence models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Mates: Model-aware data selection for efficient pretraining with data influence models

Reference 85

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.799510Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 6b82e389-4073-4ce6-9548-9f16ee90c6c5 · outbound

This paper cites Rrhf: Rank responses to align language models with human feedback.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Rrhf: Rank responses to align language models with human feedback

Reference 86

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.785266Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 1545fe36-70d1-4250-8e09-1041f3ed5f4b · outbound

This paper cites Learning to Execute.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Learning to Execute

Reference 87

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.058622Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation af7312c9-8c9e-4264-a862-3898fe373720 · outbound

This paper cites A Survey of Large Language Models.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples A Survey of Large Language Models

Reference 88

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.062800Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f423684f-2bfc-493e-9f0c-f0ec5536991d · outbound

This paper cites SLiC-HF: Sequence Likelihood Calibration with Human Feedback.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples SLiC-HF: Sequence Likelihood Calibration with Human Feedback

Reference 89

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.067249Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c5300fb5-7f8d-4383-a4df-7bea9bcb9b16 · outbound

This paper cites Judging llm-as-a-judge with mt-bench and chatbot arena.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Judging llm-as-a-judge with mt-bench and chatbot arena

Reference 90

Resolution: unresolved; no resolver link, observed 2026-08-08T11:57:12.071918Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9bf9302b-dd28-4f2a-83a4-507ded88943e · outbound

This paper cites Lima: Less is more for alignment.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Lima: Less is more for alignment

Reference 91

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.762329Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation e610697b-1579-4e58-9e90-d7d9984f5ffc · outbound

This paper cites R., Zhao, S., Song, K., Xu, S., and Zhu, C.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples R., Zhao, S., Song, K., Xu, S., and Zhu, C

Reference 92

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.746027Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 92d66173-e377-4369-bd29-6c1e14643f46 · outbound

This paper cites Iterative data smoothing: Mitigating reward overfitting and overoptimization in rlhf.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Iterative data smoothing: Mitigating reward overfitting and overoptimization in rlhf

Reference 93

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.732586Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 076af431-e5b3-4735-b8ec-b743b489b3f0 · outbound

This paper cites Exploring the learning difficulty of data: Theory and measure.

Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples Exploring the learning difficulty of data: Theory and measure

Reference 94

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T11:57:12.713528Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Pith citing papers

Observation 52494504-8160-465d-9634-fae864efedaa · inbound

CheXPO: Preference Optimization for Chest X-ray VLMs with Counterfactual Rationale cites this paper.

CheXPO: Preference Optimization for Chest X-ray VLMs with Counterfactual Rationale Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples

Reference 2025

Resolution: unresolved; no resolver link, observed 2026-08-06T18:55:17.631077Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 587ec8ef-4609-4aaa-b72b-9c19191f614d · inbound

Difficulty-Based Preference Data Selection by DPO Implicit Reward Gap cites this paper.

Difficulty-Based Preference Data Selection by DPO Implicit Reward Gap Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples

Reference 20

Resolution: verified exact; arxiv_id, observed 2026-05-21T23:50:47.602048Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 2116b96e-ee3c-472a-ad86-24fcc94936e6 · inbound

Revisiting Robustness for LLM Safety Alignment via Selective Geometry Control cites this paper.

Revisiting Robustness for LLM Safety Alignment via Selective Geometry Control Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples

Reference 8

Resolution: verified exact; arxiv_id, observed 2026-05-22T11:21:28.957156Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 920f037d-8d99-4794-8d44-df315aa8d5cd · inbound

DharmaOCR: Specialized Small Language Models for Structured OCR that outperform Open-Source and Commercial Baselines cites this paper.

DharmaOCR: Specialized Small Language Models for Structured OCR that outperform Open-Source and Commercial Baselines Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples

Reference 54

Resolution: verified exact; arxiv_id, observed 2026-05-10T13:20:25.566649Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 71147718-e90a-4306-bc05-e8a15b8b11cb · inbound

The Hidden Bias of Process Reward Models:PRISM for Rewarding the Right Reasoning cites this paper.

The Hidden Bias of Process Reward Models:PRISM for Rewarding the Right Reasoning Principled Data Selection for Alignment: The Hidden Risks of Difficult Examples

Reference 16

Resolution: verified exact; arxiv_id, observed 2026-07-03T00:37:30.067973Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.