Paper Citation Record · LEDGER

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations

As of 15 August 2026, this Paper Citation Record lists 53 of 53 outbound references and 1 inbound Pith citation observation for arXiv:2507.05644.

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

pith.paper-citation-record.v1
2507.05644 v2

Coverage vector

measured 53 of 53 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-06T19:31:23.888184Z

measured 54 of 54 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-14T06:32:32.682623+00:00

measured 1 of 1 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-06-29T22:53:30.561450Z

measured 0 of 1 external citation measurements

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

Source: arxiv_reference, observed 2026-06-29T22:54:00.682463Z

Reference resolution

53 of 53 outbound references displayed

verified exact0
verified fuzzy24
unresolved29
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation d2a7dd51-c5bb-4e64-ba91-6e51baf603be · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 1

Resolution: unresolved; raw_fallback, observed 2026-08-06T19:31:25.037471Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 8dc1e045-a237-4cdb-9d54-d6c85e0aae40 · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 2

Resolution: unresolved; raw_fallback, observed 2026-08-06T19:31:25.023591Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 1c2029c5-56d5-46d8-b07b-08b1e357365f · outbound

This paper cites Arora, N.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Arora, N

Reference 3

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:25.009845Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 7dc1aa41-710e-4d04-9812-e03eea3f77be · outbound

This paper cites Arora, N.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Arora, N

Reference 4

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.996016Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation aefc0213-27b4-476a-8c64-43c8cf7fb730 · outbound

This paper cites Arora, N.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Arora, N

Reference 5

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.982231Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 6e12cb4f-73fb-49c0-9a85-0223646c5a95 · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 6

Resolution: unresolved; raw_fallback, observed 2026-08-06T19:31:24.968565Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation fc3c05ed-b535-4bb1-bc4e-601881048331 · outbound

This paper cites Barak, B.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Barak, B

Reference 7

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.955034Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 72763c8e-e5cf-4e8f-9e15-48f13c44e7d0 · outbound

This paper cites Toward universal steering and monitoring of AI models.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Toward universal steering and monitoring of AI models

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:18.912515Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 88e5ce86-8a4e-427b-ad8d-028614980bb1 · outbound

This paper cites Mechanism of feature learning in convolutional neural networks.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Mechanism of feature learning in convolutional neural networks

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:19.028549Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b81b91af-df0c-4091-a82e-b4c499256fed · outbound

This paper cites Scaling Laws for Associative Memories.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Scaling Laws for Associative Memories

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:19.171471Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 94ed4ca1-0948-41bf-b368-c8583e1764af · outbound

This paper cites Learning Associative Memories with Gradient Descent.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Learning Associative Memories with Gradient Descent

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:19.335976Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 364be1d2-467b-4b3d-ba35-6b2fd9234541 · outbound

This paper cites Damian, J.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Damian, J

Reference 12

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.940650Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation a08c7306-3f91-4f51-9c8d-7d5767e9cbf9 · outbound

This paper cites Davis and W.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Davis and W

Reference 13

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.926141Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation ddb929c6-eeb2-4083-b98d-6709255ca34d · outbound

This paper cites The Evolution of Statistical Induction Heads: In-Context Learning Markov Chains.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations The Evolution of Statistical Induction Heads: In-Context Learning Markov Chains

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:19.725776Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2eea3257-3082-4f41-bcb0-bd63cafb9b10 · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 15

Resolution: unresolved; raw_fallback, observed 2026-08-06T19:31:24.911625Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation c4371282-5327-4e9f-90cc-ff919e938528 · outbound

This paper cites Fernandez-Delgado, E.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Fernandez-Delgado, E

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.897099Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 4c035b1c-aeb2-432e-bd61-0f68a23ebe64 · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 17

Resolution: unresolved; raw_fallback, observed 2026-08-06T19:31:24.883272Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 706d7da5-4244-4ea5-aa3d-85df53b06cf3 · outbound

This paper cites SGD and Weight Decay Secretly Minimize the Rank of Your Neural Network.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations SGD and Weight Decay Secretly Minimize the Rank of Your Neural Network

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:20.239832Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8f85cd05-c950-4fb7-8c4b-4d2df31c04a4 · outbound

This paper cites Gan and T.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Gan and T

Reference 19

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.867454Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 8b77646c-ee6e-4b99-89e7-2011278f6431 · outbound

This paper cites Grokking modular arithmetic.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Grokking modular arithmetic

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:20.404668Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation aa502f56-32bf-400a-b429-ac33cba44b8b · outbound

This paper cites Gunasekar, J.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Gunasekar, J

Reference 21

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.852195Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation cd7e5fd8-d43d-4135-8fbc-21f33ddc52a2 · outbound

This paper cites Gunasekar, B.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Gunasekar, B

Reference 22

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.837780Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 1f07f671-682f-432a-9c04-4f42f1180631 · outbound

This paper cites Neural Collapse Under MSE Loss: Proximity to and Dynamics on the Central Path.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Neural Collapse Under MSE Loss: Proximity to and Dynamics on the Central Path

Reference 23

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:20.601467Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b978ebb9-e0bd-47df-a37d-d1cf202c3168 · outbound

This paper cites Jacot, F.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Jacot, F

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.823514Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 908d1050-2f3c-4e82-8e92-abb7418a68c3 · outbound

This paper cites Ji and M.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Ji and M

Reference 25

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.808800Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 3f861106-d511-430a-8734-bee57e9b49d0 · outbound

This paper cites Ji and M.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Ji and M

Reference 26

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.794164Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 950a49b5-cdf4-445e-8e06-ded169d96f66 · outbound

This paper cites Neural Collapse: A Review on Modelling Principles and Generalization.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Neural Collapse: A Review on Modelling Principles and Generalization

Reference 27

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:20.976793Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c16ca467-e6f9-411b-9775-550facc9b7a1 · outbound

This paper cites Krizhevsky, G.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Krizhevsky, G

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:21.064475Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 59989079-aa73-4069-bec2-1dbc2a1823e9 · outbound

This paper cites Grokking as the Transition from Lazy to Rich Training Dynamics.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Grokking as the Transition from Lazy to Rich Training Dynamics

Reference 29

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:21.129096Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d29c9979-b953-4e38-a1bf-c955c02af362 · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 30

Resolution: unresolved; raw_fallback, observed 2026-08-06T19:31:24.768034Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation c5225fb2-4470-4553-b909-8b34d624a417 · outbound

This paper cites Properties of the After Kernel.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Properties of the After Kernel

Reference 31

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:21.269704Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2230e2cb-89dc-40ef-a786-ec6da92d6e98 · outbound

This paper cites Gradient Descent Maximizes the Margin of Homogeneous Neural Networks.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Gradient Descent Maximizes the Margin of Homogeneous Neural Networks

Reference 32

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:21.355070Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ec1e8e2a-d381-4382-90bc-cbca99002a97 · outbound

This paper cites Mallinar, D.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Mallinar, D

Reference 33

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.752906Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation e2ea96ef-6db6-42e7-a4ec-6f12e2b8c96e · outbound

This paper cites Marion and L.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Marion and L

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.738055Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 731e296b-ac57-4d00-8e8d-241a89762000 · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 35

Resolution: unresolved; raw_fallback, observed 2026-08-06T19:31:24.722954Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation b25d254f-e221-4b1f-9701-47174cb8b4a0 · outbound

This paper cites Feature emergence via margin maximization: case studies in algebraic tasks.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Feature emergence via margin maximization: case studies in algebraic tasks

Reference 36

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:21.714527Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a4cb04f7-868d-4719-8839-5557a2c008b6 · outbound

This paper cites Progress measures for grokking via mechanistic interpretability.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Progress measures for grokking via mechanistic interpretability

Reference 37

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:21.789558Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e6ad2f75-2630-4327-84e0-ed74ac02d5f6 · outbound

This paper cites How Transformers Learn Causal Structure with Gradient Descent.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations How Transformers Learn Causal Structure with Gradient Descent

Reference 38

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:21.883499Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ecb45ebc-fede-4a06-9a48-ddf234580563 · outbound

This paper cites In-context Learning and Induction Heads.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations In-context Learning and Induction Heads

Reference 39

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:22.053685Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0ca2b527-e84b-4955-8065-e6132fc15d43 · outbound

This paper cites Radhakrishnan, D.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Radhakrishnan, D

Reference 40

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.708689Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 24902931-396d-4ebb-ae42-dfefc601ed95 · outbound

This paper cites Radhakrishnan, M.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Radhakrishnan, M

Reference 41

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.694679Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 9a72e09f-4fe4-4ee2-98d2-82d05b4f194d · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 42

Resolution: unresolved; raw_fallback, observed 2026-08-06T19:31:24.679642Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 7bb330c8-4970-42e6-847e-da508dcfbfc8 · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 43

Resolution: unresolved; raw_fallback, observed 2026-08-06T19:31:24.665209Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation fbf637da-08f3-4cf0-b627-d1e2759742b0 · outbound

This paper cites Schölkopf.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Schölkopf

Reference 44

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.650123Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation fcaf784b-b920-43de-94da-14a76e9b50b3 · outbound

This paper cites Soudry, E.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Soudry, E

Reference 45

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.634050Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 578f007b-5beb-46ed-ae21-ba15c00ceea1 · outbound

This paper cites Stewart, F.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Stewart, F

Reference 46

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.616876Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 2fee3bd5-31e5-4c92-a55e-04e0eb049272 · outbound

This paper cites Thompson.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Thompson

Reference 47

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.600429Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 620f0ca0-1e19-4cef-863c-21cb30acd2f2 · outbound

This paper cites Woodworth, S.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Woodworth, S

Reference 48

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.586352Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation b304a0c9-3587-4c97-bdcc-9f5161999290 · outbound

This paper cites Zangrando, P.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Zangrando, P

Reference 49

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:23.296676Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation cd342bf4-0045-4985-acf7-0e0c7e1e5b85 · outbound

This paper cites an unresolved cited work.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Unresolved cited work

Reference 50

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:23.400916Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation af7e1ca8-b352-49da-a3f4-621f29fa83df · outbound

This paper cites Catapults in SGD: spikes in the training loss and their impact on generalization through feature learning.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Catapults in SGD: spikes in the training loss and their impact on generalization through feature learning

Reference 51

Resolution: unresolved; no resolver link, observed 2026-08-06T19:31:23.560693Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 706c1eac-8e9a-47dd-beec-0b0a644a6a20 · outbound

This paper cites Ziyin, I.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Ziyin, I

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.572126Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Observation 11db1100-1a4e-49fe-a9b0-9225509e8ed9 · outbound

This paper cites Ziyin, B.

The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations Ziyin, B

Reference 53

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T19:31:24.557660Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.

Pith citing papers

Observation ff9315d3-5ca8-48f3-8eec-64cd09cffed3 · inbound

Emergence via Phase Transitions: Mechanism Landscapes and Universal Convergence Across Complex Systems cites this paper.

Emergence via Phase Transitions: Mechanism Landscapes and Universal Convergence Across Complex Systems The Features at Convergence Theorem: a first-principles alternative to the Neural Feature Ansatz for how networks learn representations

Reference 7

Resolution: verified exact; arxiv_id, observed 2026-06-29T22:54:00.683913Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-14T06:32:32.682623+00:00.