Paper Citation Record · LEDGER

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection

As of 17 August 2026, this Paper Citation Record lists 19 of 19 outbound references and 1 inbound Pith citation observation for arXiv:2505.03793.

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

pith.paper-citation-record.v1
2505.03793 v3

Coverage vector

measured 19 of 19 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-16T04:44:35.337310Z

measured 20 of 20 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-17T06:30:58.91139+00:00

measured 1 of 1 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-01T06:05:53.307070Z

measured 0 of 1 external citation measurements

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

Source: cited_works

Reference resolution

19 of 19 outbound references displayed

verified exact1
verified fuzzy8
unresolved8
parse uncertain0
malformed identifier2
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 6c4527d8-5631-4987-986a-acb886c59824 · outbound

This paper cites an unresolved cited work.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Unresolved cited work

Reference 1

Resolution: unresolved; raw_fallback, observed 2026-08-16T04:44:35.697231Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation c148d042-538a-4286-b0b4-5602f981fee6 · outbound

This paper cites The proof follows from standard results on multivariate normal distributions with additional attention to transformer components.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection The proof follows from standard results on multivariate normal distributions with additional attention to transformer components

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-08-16T04:44:35.680196Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 9f31eea1-3b2d-44c7-98f8-74101bf8c5b4 · outbound

This paper cites an unresolved cited work.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Unresolved cited work

Reference 3

Resolution: unresolved; raw_fallback, observed 2026-08-16T04:44:35.615207Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 37751c3f-d307-45ef-81c6-7dbc0f67d372 · outbound

This paper cites an unresolved cited work.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Unresolved cited work

Reference 4

Resolution: unresolved; raw_fallback, observed 2026-08-16T04:44:35.664047Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 26bef7c8-6bd9-4592-8c5e-0587520633b1 · outbound

This paper cites an unresolved cited work.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Unresolved cited work

Reference 5

Resolution: unresolved; raw_fallback, observed 2026-08-16T04:44:35.578469Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 65da474a-3237-4fe1-952b-fe2d650a8873 · outbound

This paper cites Note:This proof assumes allh i are non-negative real numbers, which is aligned with the property ofhi in our bound.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Note:This proof assumes allh i are non-negative real numbers, which is aligned with the property ofhi in our bound

Reference 6

Resolution: verified fuzzy; raw_fallback, observed 2026-08-16T04:44:35.562598Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 48f3119a-d28c-4ecf-97cf-c52967010f42 · outbound

This paper cites an unresolved cited work.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Unresolved cited work

Reference 7

Resolution: unresolved; raw_fallback, observed 2026-08-16T04:44:35.648502Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 098fadc5-6705-45a1-95f2-0ddd04429776 · outbound

This paper cites an unresolved cited work.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Unresolved cited work

Reference 8

Resolution: unresolved; raw_fallback, observed 2026-08-16T04:44:35.632614Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation aaba6cb9-d809-418c-9475-f13b403098b6 · outbound

This paper cites an unresolved cited work.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Unresolved cited work

Reference 10

Resolution: unresolved; raw_fallback, observed 2026-08-16T04:44:35.595862Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation e80732c5-b3b5-47b5-b3f7-6a08a12ff5af · outbound

This paper cites • The Hessian matrix for this loss function is defined as: H=∇ 2 θL(θ)(17) For each layerl, letH l be the Hessian of the loss function with respect to the parameters in that layer.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection • The Hessian matrix for this loss function is defined as: H=∇ 2 θL(θ)(17) For each layerl, letH l be the Hessian of the loss function with respect to the parameters in that layer

Reference 13

Resolution: verified fuzzy; raw_fallback, observed 2026-08-16T04:44:35.546343Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation c96eec89-f7a0-47f7-86c4-c0ac520233de · outbound

This paper cites Proof.• Consider the empirical loss function: L(θ) = 1 n nX i=1 ℓ(θ;xi)(19) whereℓ(θ;x i)is the loss associated with samplex i.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Proof.• Consider the empirical loss function: L(θ) = 1 n nX i=1 ℓ(θ;xi)(19) whereℓ(θ;x i)is the loss associated with samplex i

Reference 14

Resolution: verified fuzzy; raw_fallback, observed 2026-08-16T04:44:35.528888Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation df8d0f0a-f9b3-40a5-a796-1f8ba875f596 · outbound

This paper cites Proof.• Let the variance of the gradient during fine-tuning beσ 2(n).

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Proof.• Let the variance of the gradient during fine-tuning beσ 2(n)

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-08-16T04:44:35.511374Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation ebbca4cf-f8c2-4928-b61e-1c226c53a583 · outbound

This paper cites Let’s give tr(H) =C 1n−β1 as the conclusion of this statement.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Let’s give tr(H) =C 1n−β1 as the conclusion of this statement

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-08-16T04:44:35.493985Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 38f796ed-756b-4eae-8c65-78a7695c14b9 · outbound

This paper cites The dimension of Wi isdi bydi−1, wheredi is the dimension of inputxi.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection The dimension of Wi isdi bydi−1, wheredi is the dimension of inputxi

Reference 17

Resolution: verified fuzzy; raw_fallback, observed 2026-08-16T04:44:35.476482Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 7a5d204e-d814-4519-8503-1ab8bbe37787 · outbound

This paper cites an unresolved cited work.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Unresolved cited work

Reference 18

Resolution: malformed identifier; raw_fallback, observed 2026-08-16T04:44:35.460062Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 4169904b-f97d-4f2c-9325-99d3e73cbee3 · outbound

This paper cites Specifically, the Pearson correlation drops from 78.14 (at average length 20) to 77.39 and 76.89 for lengths 18 and 22, respectively, while relative accuracy similarly declines.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Specifically, the Pearson correlation drops from 78.14 (at average length 20) to 77.39 and 76.89 for lengths 18 and 22, respectively, while relative accuracy similarly declines

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-08-16T04:44:35.444272Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation 0ec56dd8-4da4-41cd-ad80-6cf730e9c530 · outbound

This paper cites Mastering Long-Tail Complexity on Graphs: Characterization, Learning, and Generalization.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Mastering Long-Tail Complexity on Graphs: Characterization, Learning, and Generalization

Reference 635

Resolution: verified exact; local_arxiv, observed 2026-08-16T04:44:35.382347Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-17T06:30:58.91139+00:00.

Observation cdcddf9b-8abf-4c08-a717-4cf7fd53afb7 · outbound

This paper cites BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension

Reference 2020

Resolution: malformed identifier; no resolver link, observed 2026-08-16T04:44:35.248173Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ef43e5d4-d547-4e27-abe0-0957a83e72cf · outbound

This paper cites Less is More: Selective Layer Finetuning with SubTuning.

LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection Less is More: Selective Layer Finetuning with SubTuning

Reference 2022

Resolution: unresolved; no resolver link, observed 2026-08-16T04:44:35.242295Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation 9c45545e-a2ef-40c8-9486-a68a0ee71104 · inbound

Enough is as good as a feast: A Comprehensive Analysis of How Reinforcement Learning Mitigates Task Conflicts in LLMs cites this paper.

Enough is as good as a feast: A Comprehensive Analysis of How Reinforcement Learning Mitigates Task Conflicts in LLMs LENSLLM: Unveiling Fine-Tuning Dynamics for LLM Selection

Reference 23

Resolution: unresolved; no resolver link, observed 2026-08-01T06:05:53.307070Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.