Paper Citation Record · LEDGER

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less

As of 20 August 2026, this Paper Citation Record lists 52 of 52 outbound references and 3 inbound Pith citation observations for arXiv:2605.06654.

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

pith.paper-citation-record.v1
2605.06654 v1

Coverage vector

measured 52 of 52 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-05-08T12:00:49.127471Z

measured 55 of 55 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-20T06:33:59.587034+00:00

measured 3 of 3 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-04T04:19:38.424519Z

measured 0 of 1 external citation measurements

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

Source: pith, observed 2026-07-02T11:56:55.992114Z

Reference resolution

52 of 52 outbound references displayed

verified exact41
verified fuzzy4
unresolved2
parse uncertain0
malformed identifier0
metadata mismatch5

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation c2207734-b6f6-42fc-a872-15b9fd029f84 · outbound

This paper cites arXiv preprint arXiv:2512.16928 , year=.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less arXiv preprint arXiv:2512.16928 , year=

Reference 1

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T19:26:08.621624Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 1a26534a-2a18-4038-9238-482c3f75d6d6 · outbound

This paper cites The Geometry of Sign Gradient Descent.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less The Geometry of Sign Gradient Descent

Reference 2

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.613799Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 7f285ec0-7c51-4989-94a2-5e67b5bff8f7 · outbound

This paper cites Old Optimizer, New Norm: An Anthology.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Old Optimizer, New Norm: An Anthology

Reference 3

Resolution: verified exact; arxiv_id, observed 2026-05-16T07:27:53.041705Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation ad4e9f03-6c4e-478a-826b-f9b537b6122c · outbound

This paper cites LoRA Learns Less and Forgets Less.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less LoRA Learns Less and Forgets Less

Reference 4

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T19:26:08.668232Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 2c37b115-b48a-4f6b-83e9-cfa5e1166495 · outbound

This paper cites Why Gradients Rapidly Increase Near the End of Training.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Why Gradients Rapidly Increase Near the End of Training

Reference 5

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.465260Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 9aa17456-7f4c-4915-8828-f9adb80ba8cb · outbound

This paper cites The Llama 3 Herd of Models.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less The Llama 3 Herd of Models

Reference 6

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.588850Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 7d23d21a-41e1-4567-924e-045438fda472 · outbound

This paper cites Gaussian Error Linear Units (GELUs).

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Gaussian Error Linear Units (GELUs)

Reference 7

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.454674Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 75507299-eb8a-4078-83fb-a22a249ceb3b · outbound

This paper cites Measuring Forgetting of Memorized Training Examples.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Measuring Forgetting of Memorized Training Examples

Reference 8

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.593021Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 7ee54a71-d2e5-47f8-929b-96ac55cfdafa · outbound

This paper cites Provable Complexity Improvement of AdaGrad over SGD: Upper and Lower Bounds in Stochastic Non-Convex Optimization.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Provable Complexity Improvement of AdaGrad over SGD: Upper and Lower Bounds in Stochastic Non-Convex Optimization

Reference 9

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.606390Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 08cf0d11-a3bd-41b6-be38-ff7898efc49b · outbound

This paper cites Kimi K2.5: Visual Agentic Intelligence.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Kimi K2.5: Visual Agentic Intelligence

Reference 10

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.580519Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 36fc57b1-e1fc-445b-9dbc-effb24ae7b67 · outbound

This paper cites Adam: A Method for Stochastic Optimization.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Adam: A Method for Stochastic Optimization

Reference 11

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.584831Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation f777e85e-7854-414a-a13e-ad3cc895b8ce · outbound

This paper cites Noise Is Not the Main Factor Behind the Gap Between SGD and Adam on Transformers, but Sign Descent Might Be.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Noise Is Not the Main Factor Behind the Gap Between SGD and Adam on Transformers, but Sign Descent Might Be

Reference 12

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.565537Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 4ac87229-543a-4618-82ed-a76d92920cdc · outbound

This paper cites Muon is Scalable for LLM Training.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Muon is Scalable for LLM Training

Reference 13

Resolution: verified exact; arxiv_id, observed 2026-05-11T23:02:52.981675Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 20218c8c-a913-4081-90e6-74a40eeb76b6 · outbound

This paper cites AdaGrad under Anisotropic Smoothness.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less AdaGrad under Anisotropic Smoothness

Reference 14

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.577041Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 96edba34-090c-4e5c-b12d-8427fd30a166 · outbound

This paper cites SGDR: Stochastic Gradient Descent with Warm Restarts.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less SGDR: Stochastic Gradient Descent with Warm Restarts

Reference 15

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.573133Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 4f098732-d3b8-4da8-b2ed-62c4bafda5c2 · outbound

This paper cites Decoupled Weight Decay Regularization.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Decoupled Weight Decay Regularization

Reference 16

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.600508Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation f4527207-05ea-4dd8-9507-941d1a833de2 · outbound

This paper cites Sculpting Subspaces: Constrained Full Fine-Tuning in LLMs for Continual Learning.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Sculpting Subspaces: Constrained Full Fine-Tuning in LLMs for Continual Learning

Reference 17

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.634575Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 1e85877e-44bc-4d15-a1a8-5749c9a97cc2 · outbound

This paper cites Unbiased gradient low-rank projection.arXiv preprint arXiv:2510.17802.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Unbiased gradient low-rank projection.arXiv preprint arXiv:2510.17802

Reference 18

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.558293Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 8de20694-9b62-4668-9ea0-79d4fddf3e94 · outbound

This paper cites Training Deep Learning Models with Norm-Constrained LMOs.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Training Deep Learning Models with Norm-Constrained LMOs

Reference 19

Resolution: verified exact; arxiv_id, observed 2026-05-21T21:22:37.653978Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation f1f02a4f-4515-4026-a6d2-bd867ff84826 · outbound

This paper cites icarl: Incre- mental classifier and representation learning.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less icarl: Incre- mental classifier and representation learning

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-05-26T13:47:25.028824Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 7a950eec-14aa-4e47-ab6c-b8982a1f4b85 · outbound

This paper cites On the Convergence of Adam and Beyond.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less On the Convergence of Adam and Beyond

Reference 21

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.480366Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation e69b531a-65e8-493c-9425-606bceeda5fb · outbound

This paper cites (How) Learning Rates Regulate Catastrophic Overtraining.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less (How) Learning Rates Regulate Catastrophic Overtraining

Reference 22

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.516618Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation c241f457-713a-41f8-9d3e-7a495da4d34b · outbound

This paper cites Benchmarking Optimizers for Large Language Model Pretraining.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Benchmarking Optimizers for Large Language Model Pretraining

Reference 23

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.543311Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation fd97bb7d-23c4-4df2-b005-67030e28bb84 · outbound

This paper cites GLU Variants Improve Transformer.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less GLU Variants Improve Transformer

Reference 24

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.625847Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation ee1c7cdd-d492-4dec-a458-30197356ce85 · outbound

This paper cites Lora vs full fine-tuning: An illusion of equivalence.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Lora vs full fine-tuning: An illusion of equivalence

Reference 25

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.547227Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation b97e7a15-a4d7-4248-9177-2ad6489cbcff · outbound

This paper cites Overtrained Language Models Are Harder to Fine-Tune.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Overtrained Language Models Are Harder to Fine-Tune

Reference 26

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.539517Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation ab44dce8-7b8b-44cb-902e-4aa2bd1a0bc3 · outbound

This paper cites Less Regret via Online Conditioning.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Less Regret via Online Conditioning

Reference 27

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.508615Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 1e3575ad-ab31-4b6d-8858-e9664cfd7ddf · outbound

This paper cites ArXiv Preprint: 2511.00674 , Year =.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less ArXiv Preprint: 2511.00674 , Year =

Reference 28

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T19:26:08.610339Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation d8f36dc5-1a7a-4d95-a955-201b4894fdc2 · outbound

This paper cites Llama 2: Open Foundation and Fine-Tuned Chat Models.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Llama 2: Open Foundation and Fine-Tuned Chat Models

Reference 29

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.512185Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 29e05265-12eb-40cc-95fd-de75bd8b3692 · outbound

This paper cites SOAP: Improving and Stabilizing Shampoo using Adam.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less SOAP: Improving and Stabilizing Shampoo using Adam

Reference 30

Resolution: verified exact; arxiv_id, observed 2026-05-15T01:02:28.631723Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 6a34c718-50c3-41dc-a453-89bed6c500bc · outbound

This paper cites Muon outperforms adam in tail-end associative memory learning.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Muon outperforms adam in tail-end associative memory learning

Reference 31

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.554861Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 8347914b-4c3a-4e8e-aefe-e7d261dbc576 · outbound

This paper cites Magicoder: Empowering Code Generation with OSS-Instruct.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Magicoder: Empowering Code Generation with OSS-Instruct

Reference 32

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.655977Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation fbb0ff62-ebc5-4ca2-8e0d-a4c89742f877 · outbound

This paper cites Fantastic Pretraining Optimizers and Where to Find Them.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Fantastic Pretraining Optimizers and Where to Find Them

Reference 33

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.470520Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 804a2766-31e4-49f3-92f5-fb3a064bf02b · outbound

This paper cites Structured Preconditioners in Adaptive Optimization: A Unified Analysis.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Structured Preconditioners in Adaptive Optimization: A Unified Analysis

Reference 34

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.660083Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation ccac7eb6-8554-4f1d-ba41-a2562d30ea4f · outbound

This paper cites Controlled llm training on spectral sphere.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Controlled llm training on spectral sphere

Reference 35

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.531367Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation c4701b1e-aca3-42f1-8db0-da32010a60c0 · outbound

This paper cites On the width scaling of neural optimizers under matrix operator norms i: Row/column normalization and hyperparameter transfer.arXiv preprint arXiv:2603.09952.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less On the width scaling of neural optimizers under matrix operator norms i: Row/column normalization and hyperparameter transfer.arXiv preprint arXiv:2603.09952

Reference 36

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.475919Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 300a9d68-6a8e-4151-aa35-e275248e36ae · outbound

This paper cites Qwen3 Technical Report.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Qwen3 Technical Report

Reference 37

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.629870Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 982ace06-fbed-4f02-b303-9cc6e2c4cdfa · outbound

This paper cites A Spectral Condition for Feature Learning.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less A Spectral Condition for Feature Learning

Reference 38

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.596926Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation d01d0395-cad3-4f9b-8c96-e98959813614 · outbound

This paper cites Large Batch Optimization for Deep Learning: Training BERT in 76 minutes.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Large Batch Optimization for Deep Learning: Training BERT in 76 minutes

Reference 39

Resolution: verified exact; arxiv_id, observed 2026-05-21T21:39:00.104164Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 06f3d469-ea31-425a-b541-7c6f1cd48023 · outbound

This paper cites StoSignSGD: Unbiased Structural Stochasticity Fixes SignSGD for Training Large Language Models.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less StoSignSGD: Unbiased Structural Stochasticity Fixes SignSGD for Training Large Language Models

Reference 40

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.562081Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation f9e4fb2e-0950-4211-b854-5b50d1359a24 · outbound

This paper cites MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Models.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less MetaMath: Bootstrap Your Own Mathematical Questions for Large Language Models

Reference 41

Resolution: verified exact; arxiv_id, observed 2026-05-13T10:07:53.977132Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 538e25c1-2441-4600-8506-41ad000ad01b · outbound

This paper cites MARS: Unleashing the Power of Variance Reduction for Training Large Models.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less MARS: Unleashing the Power of Variance Reduction for Training Large Models

Reference 42

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.551157Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 7998c599-731f-4125-89ab-d025e16f3398 · outbound

This paper cites ADADELTA: An Adaptive Learning Rate Method.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less ADADELTA: An Adaptive Learning Rate Method

Reference 43

Resolution: verified exact; arxiv_id, observed 2026-05-11T19:26:08.535605Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation a21fa863-90da-41e5-9bfa-ccd8459b71d7 · outbound

This paper cites GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

Reference 44

Resolution: verified exact; local_arxiv, observed 2026-05-11T19:26:08.642431Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 2b360052-3f06-406c-91f1-0c3317a26c5a · outbound

This paper cites Understanding deep learning requires rethinking generalization.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Understanding deep learning requires rethinking generalization

Reference 45

Resolution: verified exact; arxiv_id, observed 2026-05-13T11:56:40.372790Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation b7dff74f-a846-48eb-8e76-8b915e9acffc · outbound

This paper cites Why Transformers Need Adam: A Hessian Perspective.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Why Transformers Need Adam: A Hessian Perspective

Reference 46

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T19:26:08.489778Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 9d9eb6cb-2630-4c13-a323-20ebf4406cca · outbound

This paper cites LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models

Reference 47

Resolution: metadata mismatch; local_arxiv, observed 2026-05-11T19:26:08.526331Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 34dcb554-30af-42cc-b424-357df4da4d06 · outbound

This paper cites an unresolved cited work.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Unresolved cited work

Reference 48

Resolution: unresolved; raw_fallback, observed 2026-05-26T13:47:25.031255Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 54569520-ba6a-4626-91be-97757a02b5b6 · outbound

This paper cites Here, LoRA rank 64 shows a greater forgetting compared to rank 256, mainly because rank 256 diverges for lr=5e-4, and a smaller learning rate leads to less forgetting.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Here, LoRA rank 64 shows a greater forgetting compared to rank 256, mainly because rank 256 diverges for lr=5e-4, and a smaller learning rate leads to less forgetting

Reference 49

Resolution: verified fuzzy; raw_fallback, observed 2026-05-26T13:47:25.025446Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 581d67f9-75c1-41ff-b6eb-d653bb622d9a · outbound

This paper cites an unresolved cited work.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less Unresolved cited work

Reference 50

Resolution: unresolved; raw_fallback, observed 2026-05-26T13:47:25.019542Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 2469ed4d-f60e-4a5b-8514-9601ba82e64b · outbound

This paper cites B.2 More Detailed Activation Plots In Figures 10 and 11, we present the average activation sparsity of detailed modules and specific ac- tivation splits.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less B.2 More Detailed Activation Plots In Figures 10 and 11, we present the average activation sparsity of detailed modules and specific ac- tivation splits

Reference 51

Resolution: verified fuzzy; raw_fallback, observed 2026-05-26T13:47:25.016966Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 57ef95fa-30bb-4711-b3c2-176177269826 · outbound

This paper cites 23 •Ifα∈(α 1,∞], based on Lemma 2 and Assumption 4, it holds that ∥∆W∥ α1,β∗ ≤ ∥∆W∥ α,β∗ andE h ∥x∥2 α1 i = Θ E h ∥x∥2 α i.

Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less 23 •Ifα∈(α 1,∞], based on Lemma 2 and Assumption 4, it holds that ∥∆W∥ α1,β∗ ≤ ∥∆W∥ α,β∗ andE h ∥x∥2 α1 i = Θ E h ∥x∥2 α i

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-05-26T13:47:25.022703Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Pith citing papers

Observation 0dcd802d-c6f2-4675-a64f-0c12f2d4c537 · inbound

Double Preconditioning (DoPr): Optimization for Test-Time Performance, not Validation Loss cites this paper.

Double Preconditioning (DoPr): Optimization for Test-Time Performance, not Validation Loss Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less

Reference 185

Resolution: metadata mismatch; local_arxiv, observed 2026-07-02T11:56:55.993275Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-20T06:33:59.587034+00:00.

Observation 259292d8-88cf-4435-a378-1f4e34a0f944 · inbound

When Does Muon Help Agentic Reinforcement Learning? cites this paper.

When Does Muon Help Agentic Reinforcement Learning? Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-01T21:13:39.201865Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 31f5b1af-31b0-4aac-9ee3-6833d9bf061e · inbound

When Does Muon Help Agentic Reinforcement Learning? cites this paper.

When Does Muon Help Agentic Reinforcement Learning? Optimizer-Model Consistency: Full Finetuning with the Same Optimizer as Pretraining Forgets Less

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-04T04:19:38.424519Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.