Paper Citation Record · LEDGER

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions

As of 21 August 2026, this Paper Citation Record lists 73 of 73 outbound references and 2 inbound Pith citation observations for arXiv:2506.13234.

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

pith.paper-citation-record.v1
2506.13234 v1

Coverage vector

measured 73 of 73 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-07T00:43:09.868839Z

measured 75 of 75 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 2 of 2 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-01T14:42:34.657120Z

measured 0 of 1 external citation measurements

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

Source: arxiv_reference, observed 2026-07-04T10:39:45.033475Z

Reference resolution

73 of 73 outbound references displayed

verified exact3
verified fuzzy48
unresolved22
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation eb4a1a6b-df93-45ea-9a23-44c197a959d6 · outbound

This paper cites write newline.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions write newline

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:05.100047Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9455d820-76eb-4d57-a063-c9bc37aa954e · outbound

This paper cites Layer-wise linear mode connectivity.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Layer-wise linear mode connectivity

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.442863Z

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 0d47d3df-3b99-495c-8745-4434ed4d09ad · outbound

This paper cites Git re-basin: Merging models modulo permutation symmetries.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Git re-basin: Merging models modulo permutation symmetries

Reference 3

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.430227Z

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 33fc3d64-5ff8-4ce7-beb1-416cb4bf9be0 · outbound

This paper cites S., Bachmann, G., Noci, L., and Hofmann, T.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions S., Bachmann, G., Noci, L., and Hofmann, T

Reference 4

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.416386Z

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 f07bd3ba-185b-4ae8-a73b-9f5d1d0f2f88 · outbound

This paper cites Layer Normalization.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Layer Normalization

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:05.358766Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2ea6b26c-e97e-4067-b849-f42b9649ae33 · outbound

This paper cites V., Akram, Y., Zucchet, N., Aitchison, L., and Steger, A.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions V., Akram, Y., Zucchet, N., Aitchison, L., and Steger, A

Reference 6

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.403319Z

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 f2dd72fa-b764-4325-81f2-c213f5ffdd3b · outbound

This paper cites Shift-Curvature, SGD, and Generalization.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Shift-Curvature, SGD, and Generalization

Reference 7

Resolution: verified exact; local_arxiv, observed 2026-08-07T00:43:10.802107Z

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 d57522d1-d45e-4457-93bc-0d847f0af79f · outbound

This paper cites Training Verifiers to Solve Math Word Problems.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Training Verifiers to Solve Math Word Problems

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:05.652763Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6deacaab-fc79-46a1-8190-8030d5fbd5e1 · outbound

This paper cites Gradient Descent on Neural Networks Typically Occurs at the Edge of Stability.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Gradient Descent on Neural Networks Typically Occurs at the Edge of Stability

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:05.706153Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cae1a793-4f5b-45d6-94d1-29bbed05d785 · outbound

This paper cites Why Do We Need Weight Decay in Modern Deep Learning?.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Why Do We Need Weight Decay in Modern Deep Learning?

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:05.785980Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 77d0ed8a-39b4-47e0-8a28-996c0917c820 · outbound

This paper cites BERT : Pre-training of deep bidirectional transformers for language understanding.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions BERT : Pre-training of deep bidirectional transformers for language understanding

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:05.840133Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 97ec0c4d-026f-4e71-959b-c8c5f9192cbd · outbound

This paper cites An image is worth 16x16 words: Transformers for image recognition at scale.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions An image is worth 16x16 words: Transformers for image recognition at scale

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:05.927440Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5576d2a3-c04d-4dbd-9544-f334136af34a · outbound

This paper cites Essentially No Barriers in Neural Network Energy Landscape.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Essentially No Barriers in Neural Network Energy Landscape

Reference 14

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.380962Z

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 ecd9b941-8d30-47ff-9d7b-3a26ea28d232 · outbound

This paper cites The role of permutation invariance in linear mode connectivity of neural networks.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions The role of permutation invariance in linear mode connectivity of neural networks

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.368385Z

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 bc154909-b8b3-4591-841b-f9bde4410cbb · outbound

This paper cites Deep ensembles: A loss landscape perspective, 2019.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Deep ensembles: A loss landscape perspective, 2019

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.355836Z

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 c1c54ca5-d3e2-48ed-8b50-aaea4eec6b04 · outbound

This paper cites K., Paul, M., Kharaghani, S., Roy, D.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions K., Paul, M., Kharaghani, S., Roy, D

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:06.167369Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a2b4c10b-c885-4911-ba48-7379fee3a0ba · outbound

This paper cites and Carbin, M.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions and Carbin, M

Reference 18

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.342084Z

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 ea1184bf-b8c5-4632-a068-56b4db380d37 · outbound

This paper cites K., Roy, D., and Carbin, M.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions K., Roy, D., and Carbin, M

Reference 19

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.329733Z

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 2c9a9f62-50f6-4083-9476-8159135a0b5f · outbound

This paper cites J., and Morcos, A.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions J., and Morcos, A

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.317064Z

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 36d495ba-d25d-4641-831e-2804117bc2c3 · outbound

This paper cites P., and Wilson, A.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions P., and Wilson, A

Reference 21

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.302005Z

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 bd56b550-3ce3-463a-a077-a0e3fa02fde6 · outbound

This paper cites Qualitatively characterizing neural network optimization problems.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Qualitatively characterizing neural network optimization problems

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:06.474225Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 20f28c73-28f5-4741-a562-6a58337b62a8 · outbound

This paper cites OLMo: Accelerating the Science of Language Models.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions OLMo: Accelerating the Science of Language Models

Reference 23

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:06.553941Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0e32780e-c0f6-4fa8-a367-b5b479cc4613 · outbound

This paper cites Delving deep into rectifiers: Surpassing human-level performance on imagenet classification.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Delving deep into rectifiers: Surpassing human-level performance on imagenet classification

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.289431Z

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 09cab4be-4db1-47c5-a25a-01b62d84faa9 · outbound

This paper cites Deep residual learning for image recognition.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Deep residual learning for image recognition

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:06.686925Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 25a5439d-49ea-4a25-a5bb-4fcd7389b0e4 · outbound

This paper cites E., and Weinberger, K.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions E., and Weinberger, K

Reference 26

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.267568Z

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 03d99161-8650-45e4-833e-9b88a1bfdf8d · outbound

This paper cites T., Wortsman, M., Schmidt, L., Hajishirzi, H., and Farhadi, A.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions T., Wortsman, M., Schmidt, L., Hajishirzi, H., and Farhadi, A

Reference 27

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.255279Z

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 417228dd-0e58-4cf0-b78c-469f8fdcd366 · outbound

This paper cites Analysis of linear mode connectivity via permutation-based weight matching.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Analysis of linear mode connectivity via permutation-based weight matching

Reference 28

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.243071Z

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 cabbe003-5788-42ce-9442-0e06f1dc2a04 · outbound

This paper cites Maximal initial learning rates in deep R e LU networks.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Maximal initial learning rates in deep R e LU networks

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.231249Z

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 3a9f5bda-6533-471a-80e1-37daab5a6d86 · outbound

This paper cites Averaging weights leads to wider optima and better generalization.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Averaging weights leads to wider optima and better generalization

Reference 30

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.219123Z

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 e7dfa132-d921-48d4-9040-ad169610af87 · outbound

This paper cites Neural tangent kernel: Convergence and generalization in neural networks.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Neural tangent kernel: Convergence and generalization in neural networks

Reference 31

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.207403Z

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 91ec5cfe-c0a0-4f4d-a628-93341303a18c · outbound

This paper cites The break-even point on optimization trajectories of deep neural networks.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions The break-even point on optimization trajectories of deep neural networks

Reference 32

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.194184Z

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 5680b034-4ad2-4ca5-a62c-8d22557b3b94 · outbound

This paper cites REPAIR : RE normalizing P ermuted A ctivations for I nterpolation R epair.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions REPAIR : RE normalizing P ermuted A ctivations for I nterpolation R epair

Reference 33

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.182327Z

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 ffe8922c-d01a-4f71-8fc0-17e1916f52f6 · outbound

This paper cites Linear connectivity reveals generalization strategies.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Linear connectivity reveals generalization strategies

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.170161Z

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 a222b76a-489c-4bab-8efa-d1d941d2813c · outbound

This paper cites S., Mudigere, D., Nocedal, J., Smelyanskiy, M., and Tang, P.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions S., Mudigere, D., Nocedal, J., Smelyanskiy, M., and Tang, P

Reference 35

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.155578Z

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 d45fd1e2-f2c3-4175-8297-26281d2fcfa1 · outbound

This paper cites Similarity of neural network representations revisited.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Similarity of neural network representations revisited

Reference 36

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.143117Z

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 8eac8671-685a-42b2-9939-515ecee88f0d · outbound

This paper cites Learning multiple layers of features from tiny images, 2009.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Learning multiple layers of features from tiny images, 2009

Reference 37

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.131122Z

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 1e8b9c71-3531-484c-8fe4-eb7c2113075b · outbound

This paper cites D., Kwok, D., Matelsky, J.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions D., Kwok, D., Matelsky, J

Reference 38

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.118603Z

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 77247a5f-663b-4c50-9df5-5cdbafd649b1 · outbound

This paper cites Wide neural networks of any depth evolve as linear models under gradient descent.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Wide neural networks of any depth evolve as linear models under gradient descent

Reference 39

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.106540Z

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 e4804cfb-3121-4c0b-8e13-09f1381aeb03 · outbound

This paper cites Exploring Neural Network Landscapes: Star-Shaped and Geodesic Connectivity.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Exploring Neural Network Landscapes: Star-Shaped and Geodesic Connectivity

Reference 40

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:07.446769Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 52ca1ad0-e19e-4a3a-ac89-c5e8fbcaf0ae · outbound

This paper cites How good is a single basin? In Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, volume 238, pp.\ 4015--4023.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions How good is a single basin? In Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, volume 238, pp.\ 4015--4023

Reference 41

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.094633Z

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 ccce7def-b85f-42b9-9545-078673ddb725 · outbound

This paper cites Decoupled Weight Decay Regularization.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Decoupled Weight Decay Regularization

Reference 42

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:07.536563Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e09601d7-8bf7-4ed0-a19e-22621c0ef130 · outbound

This paper cites S., Bigelow, E.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions S., Bigelow, E

Reference 43

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.081529Z

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 d530d861-cfa0-4c6b-8b64-ff174e2c618f · outbound

This paper cites I., Farajtabar, M., Gorur, D., Pascanu, R., and Ghasemzadeh, H.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions I., Farajtabar, M., Gorur, D., Pascanu, R., and Ghasemzadeh, H

Reference 44

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.067325Z

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 5743837a-51b9-4183-b8fc-56d03a1edcd6 · outbound

This paper cites Equivariant deep weight space alignment.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Equivariant deep weight space alignment

Reference 45

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.049517Z

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 74629502-a691-494e-ab26-0172285ff859 · outbound

This paper cites What is being transferred in transfer learning? In Advances in Neural Information Processing Systems, volume 33, pp.\ 512--523, 2020.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions What is being transferred in transfer learning? In Advances in Neural Information Processing Systems, volume 33, pp.\ 512--523, 2020

Reference 46

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.036674Z

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 dc310133-757f-4c46-9bf9-c20e61825e91 · outbound

This paper cites Do wide and deep networks learn the same things? Uncovering how neural network representations vary with width and depth.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Do wide and deep networks learn the same things? Uncovering how neural network representations vary with width and depth

Reference 47

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:13.023113Z

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 01cbc6ee-6236-4bb2-82b0-5fe47f372231 · outbound

This paper cites an unresolved cited work.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Unresolved cited work

Reference 48

Resolution: unresolved; raw_fallback, observed 2026-08-07T00:43:13.009630Z

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 b2a2efc0-63fc-4835-b443-bed057b68b29 · outbound

This paper cites SVCCA : Singular vector canonical correlation analysis for deep learning dynamics and interpretability.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions SVCCA : Singular vector canonical correlation analysis for deep learning dynamics and interpretability

Reference 49

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.995802Z

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 1cd3e859-ba27-4f3c-a226-9d4acaccb2e0 · outbound

This paper cites T., Bello-Rivas, J.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions T., Bello-Rivas, J

Reference 50

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.984018Z

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 c7eb8cbe-1ba9-467a-8418-e42f50c2d1a2 · outbound

This paper cites C., and Fei-Fei, L.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions C., and Fei-Fei, L

Reference 51

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:07.968408Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 42b97185-5501-4073-a8d4-7fa89fa81198 · outbound

This paper cites P., and Lobacheva, E.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions P., and Lobacheva, E

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.972028Z

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 1053087f-ff0e-4a9c-be33-be7f95d33d93 · outbound

This paper cites Random teachers are good teachers.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Random teachers are good teachers

Reference 53

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.960005Z

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 2e255c62-1ab0-4453-a5c7-db42754a467a · outbound

This paper cites The MultiBERTs : BERT reproductions for robustness analysis.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions The MultiBERTs : BERT reproductions for robustness analysis

Reference 54

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.946040Z

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 5ed3dca2-8616-4a1b-b413-4c8f48e93571 · outbound

This paper cites M., Rolnick, D., and Dziugaite, G.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions M., Rolnick, D., and Dziugaite, G

Reference 55

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.932301Z

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 a588ae60-8204-425c-95ee-ea988037edef · outbound

This paper cites Geometry of the loss landscape in overparameterized neural networks: Symmetries and invariances.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Geometry of the loss landscape in overparameterized neural networks: Symmetries and invariances

Reference 56

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.920243Z

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 c654e2c6-6839-425e-a11b-fe6189ad625d · outbound

This paper cites an unresolved cited work.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Unresolved cited work

Reference 57

Resolution: unresolved; raw_fallback, observed 2026-08-07T00:43:12.887670Z

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 984c3f57-b918-4bca-ad22-68600c4ececd · outbound

This paper cites P., Adilova, L., Kamp, M., Fischer, A., Schölkopf, B., Tübingen, M.-I., Hofmann, T., and Ch, E.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions P., Adilova, L., Kamp, M., Fischer, A., Schölkopf, B., Tübingen, M.-I., Hofmann, T., and Ch, E

Reference 58

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.722674Z

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 b198fd57-bda2-4a97-9fb8-de7ac37083fe · outbound

This paper cites A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decay.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions A disciplined approach to neural network hyper-parameters: Part 1 -- learning rate, batch size, momentum, and weight decay

Reference 59

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:08.528757Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2989b5ac-13d6-4cfa-961f-be28b8d640d5 · outbound

This paper cites an unresolved cited work.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Unresolved cited work

Reference 60

Resolution: unresolved; raw_fallback, observed 2026-08-07T00:43:12.557066Z

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 ac025386-8e78-47e3-b420-663ccce4141c · outbound

This paper cites The boundary of neural network trainability is fractal.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions The boundary of neural network trainability is fractal

Reference 61

Resolution: verified exact; local_arxiv, observed 2026-08-07T00:43:10.423646Z

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 a05d770b-7108-4b4c-9d38-e3205fdff847 · outbound

This paper cites Do Deep Neural Network Solutions Form a Star Domain?.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Do Deep Neural Network Solutions Form a Star Domain?

Reference 62

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:08.736723Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0e75bfbd-d18c-48e7-901c-3750d973c91e · outbound

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

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Overtrained Language Models Are Harder to Fine-Tune

Reference 63

Resolution: unresolved; no resolver link, observed 2026-08-07T00:43:08.795214Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 73027bf7-1fbd-4c1b-83dd-a81844519b0d · outbound

This paper cites Nonlinear dynamics and chaos: with applications to physics, biology, chemistry, and engineering.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Nonlinear dynamics and chaos: with applications to physics, biology, chemistry, and engineering

Reference 64

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.412655Z

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 e110fd5b-ff1b-4c39-aa9d-0ccabe5f9352 · outbound

This paper cites W., Thiery, A.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions W., Thiery, A

Reference 65

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.261679Z

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 f13f49b5-fcd5-4786-a3b2-f6a38ac3d910 · outbound

This paper cites Weight averaging for neural networks and local resampling schemes.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Weight averaging for neural networks and local resampling schemes

Reference 66

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:12.171905Z

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 8f2e119f-7991-470a-8063-67d83a6802af · outbound

This paper cites an unresolved cited work.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Unresolved cited work

Reference 67

Resolution: unresolved; raw_fallback, observed 2026-08-07T00:43:11.934338Z

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 088c81ff-4834-4b17-a765-a5456a330f94 · outbound

This paper cites an unresolved cited work.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Unresolved cited work

Reference 68

Resolution: unresolved; raw_fallback, observed 2026-08-07T00:43:11.799216Z

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 76977578-87c1-4165-84a8-62453b27abc3 · outbound

This paper cites Federated learning with matched averaging.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Federated learning with matched averaging

Reference 69

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:11.617288Z

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 feeb1d3b-582d-4614-aecf-b5d49675d1da · outbound

This paper cites H., Kunz, E., Kornblith, S., and Linderman, S.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions H., Kunz, E., Kornblith, S., and Linderman, S

Reference 70

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:11.383302Z

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 0ec77d66-531f-43b0-be75-bbf4cb18a167 · outbound

This paper cites C., Guestrin, C., Farhadi, A., and Rastegari, M.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions C., Guestrin, C., Farhadi, A., and Rastegari, M

Reference 71

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:11.210564Z

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 4ee66f01-88a7-4531-894f-5f08e7385fb4 · outbound

This paper cites How SGD selects the global minima in over-parameterized learning: A dynamical stability perspective.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions How SGD selects the global minima in over-parameterized learning: A dynamical stability perspective

Reference 72

Resolution: verified exact; raw_fallback, observed 2026-08-07T00:43:10.236794Z

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 cdefd4ab-d017-4ebd-927d-24a6b0888673 · outbound

This paper cites K., Savarese, P.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions K., Savarese, P

Reference 73

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:11.068269Z

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 59fa170a-956e-4fdd-bfc4-b2f3452479b8 · outbound

This paper cites Going beyond linear mode connectivity: The layerwise linear feature connectivity.

The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions Going beyond linear mode connectivity: The layerwise linear feature connectivity

Reference 74

Resolution: verified fuzzy; raw_fallback, observed 2026-08-07T00:43:10.984370Z

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 5addd247-0f30-4837-865d-ac08e2016a70 · inbound

Scaling Linear Mode Connectivity and Merging to Billion Parameter Pretrained Transformers cites this paper.

Scaling Linear Mode Connectivity and Merging to Billion Parameter Pretrained Transformers The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions

Reference 10

Resolution: verified exact; arxiv_id, observed 2026-07-04T10:39:45.035541Z

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 fc7d40c4-5438-430d-b000-d6116a2985c9 · inbound

Quasi-Monte Carlo Initialization for Meta-Reinforcement Learning cites this paper.

Quasi-Monte Carlo Initialization for Meta-Reinforcement Learning The Butterfly Effect: Neural Network Training Trajectories Are Highly Sensitive to Initial Conditions

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-01T14:42:34.657120Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.