Paper Citation Record · LEDGER

Sparse Autoencoders Do Not Find Canonical Units of Analysis

As of 9 August 2026, this Paper Citation Record lists 30 of 30 outbound references and 23 inbound Pith citation observations for arXiv:2502.04878.

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

pith.paper-citation-record.v1
2502.04878 v1

Coverage vector

measured 30 of 30 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-08T21:12:13.917656Z

measured 53 of 53 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-09T06:31:02.800959+00:00

measured 23 of 23 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-07T14:03:00.843301Z

measured 1 of 1 external citation measurements

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

Source: arxiv_reference, observed 2026-08-05T02:28:24.338817Z

Reference resolution

30 of 30 outbound references displayed

verified exact0
verified fuzzy10
unresolved20
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

0
arxiv_reference, observed 2026-08-05T02:28:24.338817Z

Outbound references

Observation 874c877f-280d-497d-bf95-e3984e979e07 · outbound

This paper cites Transcoders Find Interpretable LLM Feature Circuits.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Transcoders Find Interpretable LLM Feature Circuits

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.815125Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 239891e1-7ec4-42f5-adef-30863cf8261e · outbound

This paper cites Scaling and evaluating sparse autoencoders.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Scaling and evaluating sparse autoencoders

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.828320Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1d22a8e1-406f-4e90-b0c1-ff1f72e12c7a · outbound

This paper cites Dissecting Recall of Factual Associations in Auto-Regressive Language Models.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Dissecting Recall of Factual Associations in Auto-Regressive Language Models

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.831971Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 42ee1fd8-7de2-4baf-b442-6783740ff9ee · outbound

This paper cites Language Models Represent Space and Time.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Language Models Represent Space and Time

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.835885Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3f60d5c5-bb85-431f-a961-bef702da7580 · outbound

This paper cites Interpreting Attention Layer Outputs with Sparse Autoencoders.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Interpreting Attention Layer Outputs with Sparse Autoencoders

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.839743Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3815e31b-4fd3-4f2d-84e6-28efbfb6b15c · outbound

This paper cites Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.846998Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bc6ddbfc-4f97-4441-a16a-813e4a5fcc4a · outbound

This paper cites Towards Principled Evaluations of Sparse Autoencoders for Interpretability and Control.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Towards Principled Evaluations of Sparse Autoencoders for Interpretability and Control

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.850571Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d9ba40f5-e078-4ae5-a2de-b448ab61c189 · outbound

This paper cites Locating and editing factual associations in gpt.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Locating and editing factual associations in gpt

Reference 14

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.231281Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 2e5bcfbd-e44b-40ce-8d4d-92ad719e56e0 · outbound

This paper cites Progress measures for grokking via mechanistic interpretability.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Progress measures for grokking via mechanistic interpretability

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.861544Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 49d29ef5-c3b9-4962-a2b7-4dcc3cb0f65e · outbound

This paper cites However, in practice we find that the bdecs of SAEs trained on the same latents are very similar (minimum cosine similarity of 0.9970, differing by less than 0.1% in magnitude).

Sparse Autoencoders Do Not Find Canonical Units of Analysis However, in practice we find that the bdecs of SAEs trained on the same latents are very similar (minimum cosine similarity of 0.9970, differing by less than 0.1% in magnitude)

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.167020Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation ef503a3c-c93f-4473-9a87-b88950e81bfc · outbound

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

Sparse Autoencoders Do Not Find Canonical Units of Analysis Gemma 2: Improving Open Language Models at a Practical Size

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.874504Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2ab72c83-bf32-4747-89cd-887135d947a0 · outbound

This paper cites Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.878104Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6e21bddf-f330-44fe-8cb8-15dc0fc2d78f · outbound

This paper cites Relational Composition in Neural Networks: A Survey and Call to Action.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Relational Composition in Neural Networks: A Survey and Call to Action

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.881826Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2703105c-501b-43ed-8254-cd148002a0e0 · outbound

This paper cites Typically measured as average L0 across a batch: L0 = 1 n P i ∥f (xi)∥0.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Typically measured as average L0 across a batch: L0 = 1 n P i ∥f (xi)∥0

Reference 21

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.221640Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 6b01bdb2-aa42-4233-967c-8e11c63c0211 · outbound

This paper cites This provides a gradient for training unlike the L0-norm, but suppresses latent activations harming reconstruction performance (Rajamanoharan et al., 2024a).

Sparse Autoencoders Do Not Find Canonical Units of Analysis This provides a gradient for training unlike the L0-norm, but suppresses latent activations harming reconstruction performance (Rajamanoharan et al., 2024a)

Reference 22

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.211419Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 75b836a2-3e6f-43ba-81ea-97bafebf99b6 · outbound

This paper cites make sure.

Sparse Autoencoders Do Not Find Canonical Units of Analysis make sure

Reference 23

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.200710Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 5b66ee5d-585e-48f4-ad4c-3bbdc2b858b6 · outbound

This paper cites A.5 O PEN SOURCE SAE WEIGHTS All GPT-2 Small SAEs were trained on the layer 8 residual stream, which was chosen in line with Gao et al.

Sparse Autoencoders Do Not Find Canonical Units of Analysis A.5 O PEN SOURCE SAE WEIGHTS All GPT-2 Small SAEs were trained on the layer 8 residual stream, which was chosen in line with Gao et al

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.190068Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 192fd351-476e-47c5-9b4c-101350f33c37 · outbound

This paper cites We used the TransformerLens ( https://transformerlensorg.github.io/ TransformerLens/) implementations of GPT-2 and Gemma 2 2B.

Sparse Autoencoders Do Not Find Canonical Units of Analysis We used the TransformerLens ( https://transformerlensorg.github.io/ TransformerLens/) implementations of GPT-2 and Gemma 2 2B

Reference 25

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.178619Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 4033a41d-75f9-4879-bd62-79ba8f998c96 · outbound

This paper cites an unresolved cited work.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Unresolved cited work

Reference 28

Resolution: unresolved; raw_fallback, observed 2026-08-08T21:12:14.145630Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 730dedff-5172-4142-a060-29ca0ecbe48a · outbound

This paper cites Figure 21: Sparse probing evaluation accuracy by GPT-2 SAE dictionary size across 8 benchmark datasets, with a sparse probe using the top latent.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Figure 21: Sparse probing evaluation accuracy by GPT-2 SAE dictionary size across 8 benchmark datasets, with a sparse probe using the top latent

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.134374Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 9bf66e8c-47ce-4693-8b75-852b380bbea2 · outbound

This paper cites 22 Published as a conference paper at ICLR 2025 Figure 23: TPP evaluation accuracy by GPT-2 SAE dictionary size across 2 benchmark datasets, ablating up to 50 latents.

Sparse Autoencoders Do Not Find Canonical Units of Analysis 22 Published as a conference paper at ICLR 2025 Figure 23: TPP evaluation accuracy by GPT-2 SAE dictionary size across 2 benchmark datasets, ablating up to 50 latents

Reference 30

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.122464Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 77c682d5-d3a9-459b-8f49-dee665636ed8 · outbound

This paper cites We find a lower threshold for distinguishing novel features from reconstruction features (0.4).

Sparse Autoencoders Do Not Find Canonical Units of Analysis We find a lower threshold for distinguishing novel features from reconstruction features (0.4)

Reference 41

Resolution: verified fuzzy; raw_fallback, observed 2026-08-08T21:12:14.156015Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 78cdf7f9-3d16-4384-ac2e-7458f9175793 · outbound

This paper cites In-context Learning and Induction Heads.

Sparse Autoencoders Do Not Find Canonical Units of Analysis In-context Learning and Induction Heads

Reference 1997

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.865814Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 11610842-c7ca-4265-99e0-90cde01e776d · outbound

This paper cites k-Sparse Autoencoders.

Sparse Autoencoders Do Not Find Canonical Units of Analysis k-Sparse Autoencoders

Reference 2014

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.854776Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d5f1cc76-a6a5-46fd-b5e3-13ef2965bb31 · outbound

This paper cites Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Task

Reference 2015

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.843304Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 409d70f5-73cc-4304-8ba0-9bde432a44d0 · outbound

This paper cites Improving Dictionary Learning with Gated Sparse Autoencoders.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Improving Dictionary Learning with Gated Sparse Autoencoders

Reference 2019

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.870041Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 05650c7c-ca77-4ece-abde-b43b5a1f9dd9 · outbound

This paper cites Toy Models of Superposition.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Toy Models of Superposition

Reference 2021

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.820172Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 53139257-4cc2-4178-a89a-e7ad60874756 · outbound

This paper cites Not All Language Model Features Are One-Dimensionally Linear.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Not All Language Model Features Are One-Dimensionally Linear

Reference 2022

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.824293Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5fe28006-0be0-4138-b8f7-57db0079805f · outbound

This paper cites BatchTopK Sparse Autoencoders.

Sparse Autoencoders Do Not Find Canonical Units of Analysis BatchTopK Sparse Autoencoders

Reference 2023

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.806500Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d0916881-ecaa-4c01-bb3a-c50a8aa1698f · outbound

This paper cites Sparse Autoencoders Find Highly Interpretable Features in Language Models.

Sparse Autoencoders Do Not Find Canonical Units of Analysis Sparse Autoencoders Find Highly Interpretable Features in Language Models

Reference 2024

Resolution: unresolved; no resolver link, observed 2026-08-08T21:12:13.811079Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation 9472c785-10ea-4cc4-bee1-c5ab6b941e6d · inbound

Position: Mechanistic Interpretability Should Prioritize Feature Consistency in SAEs cites this paper.

Position: Mechanistic Interpretability Should Prioritize Feature Consistency in SAEs Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-07T14:03:00.843301Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 50f4f490-74c5-4fce-b7d5-385ecd300fd4 · inbound

Stochastic Parameter Decomposition cites this paper.

Stochastic Parameter Decomposition Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-06T22:48:37.411622Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 158f39e2-2786-4adc-92dc-0e2a90c0ce6f · inbound

Teach Old SAEs New Domain Tricks with Boosting cites this paper.

Teach Old SAEs New Domain Tricks with Boosting Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 2025

Resolution: unresolved; no resolver link, observed 2026-08-06T16:40:29.030274Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 899a1934-92c6-408c-95e4-b7e8a9da13b8 · inbound

Making Interpretable Discoveries from Unstructured Data: A High-Dimensional Multiple Hypothesis Testing Approach cites this paper.

Making Interpretable Discoveries from Unstructured Data: A High-Dimensional Multiple Hypothesis Testing Approach Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 36

Resolution: metadata mismatch; arxiv_id, observed 2026-05-18T02:00:40.006102Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation f241173c-da4b-4c12-9b56-37ad8c59b6ac · inbound

Making Interpretable Discoveries from Unstructured Data: A High-Dimensional Multiple Hypothesis Testing Approach cites this paper.

Making Interpretable Discoveries from Unstructured Data: A High-Dimensional Multiple Hypothesis Testing Approach Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 36

Resolution: unresolved; no resolver link, observed 2026-08-04T00:23:28.785151Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fb0f0e78-985c-4cfb-9741-7beb6e1dfbe4 · inbound

Superposition Without Interference? Towards Isolated Interventions via Almost Orthogonal Features in Language Models cites this paper.

Superposition Without Interference? Towards Isolated Interventions via Almost Orthogonal Features in Language Models Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 2021

Resolution: unresolved; no resolver link, observed 2026-08-03T04:36:01.693349Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 383309bf-692f-4e12-bf36-4bc3435bf553 · inbound

Superposition Without Interference? Towards Isolated Interventions via Almost Orthogonal Features in Language Models cites this paper.

Superposition Without Interference? Towards Isolated Interventions via Almost Orthogonal Features in Language Models Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 2021

Resolution: unresolved; no resolver link, observed 2026-08-04T06:15:55.771792Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7048f3de-b19f-4ae9-af2e-5389e9cd3515 · inbound

Tree SAE: Learning Hierarchical Feature Structures in Sparse Autoencoders cites this paper.

Tree SAE: Learning Hierarchical Feature Structures in Sparse Autoencoders Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 13

Resolution: verified exact; arxiv_id, observed 2026-05-11T03:15:54.363637Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 7f31fb58-494e-45ad-bc04-5f179e4536d7 · inbound

Tree SAE: Learning Hierarchical Feature Structures in Sparse Autoencoders cites this paper.

Tree SAE: Learning Hierarchical Feature Structures in Sparse Autoencoders Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 13

Resolution: verified exact; arxiv_id, observed 2026-05-12T07:16:25.329793Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation dab9fcd5-3d2e-4065-85c4-0899531f3624 · inbound

Position: Mechanistic Interpretability Must Disclose Identification Assumptions for Causal Claims cites this paper.

Position: Mechanistic Interpretability Must Disclose Identification Assumptions for Causal Claims Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 13

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T02:45:58.161730Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 1fecbe30-0288-4573-b4b9-ffb45395717b · inbound

Disentangled Sparse Representations for Concept-Separated Diffusion Unlearning cites this paper.

Disentangled Sparse Representations for Concept-Separated Diffusion Unlearning Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 24

Resolution: metadata mismatch; arxiv_id, observed 2026-05-13T05:52:22.462659Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 636848fb-09da-4f38-bd53-68091360e0b3 · inbound

WriteSAE: Sparse Autoencoders for Recurrent State cites this paper.

WriteSAE: Sparse Autoencoders for Recurrent State Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 94

Resolution: verified exact; arxiv_id, observed 2026-05-14T20:59:28.784575Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation ccd7079f-fd64-42cb-adf8-1bdc9e4c1493 · inbound

WriteSAE: Sparse Autoencoders for Recurrent State cites this paper.

WriteSAE: Sparse Autoencoders for Recurrent State Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 94

Resolution: verified exact; arxiv_id, observed 2026-05-15T04:59:45.209373Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 601d00b0-efa3-4301-95ad-071b3edf7a0b · inbound

WriteSAE: Sparse Autoencoders for Recurrent State cites this paper.

WriteSAE: Sparse Autoencoders for Recurrent State Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 94

Resolution: verified exact; arxiv_id, observed 2026-05-20T21:53:47.190627Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 0a3b2eb0-d68d-43ae-a24c-623bd614bbb4 · inbound

Descriptive Collision in Sparse Autoencoder Auto-Interpretability: When One Explanation Describes Many Features cites this paper.

Descriptive Collision in Sparse Autoencoder Auto-Interpretability: When One Explanation Describes Many Features Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 16

Resolution: verified exact; arxiv_id, observed 2026-05-14T20:27:58.974178Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 11339158-52c7-4c72-86ab-7a76859a9cf6 · inbound

Sign-Aware Gated Sparse Autoencoders: Modeling Anticorrelated Features with Bi-Jump-ReLU Activations cites this paper.

Sign-Aware Gated Sparse Autoencoders: Modeling Anticorrelated Features with Bi-Jump-ReLU Activations Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 22

Resolution: metadata mismatch; arxiv_id, observed 2026-06-29T14:23:30.750358Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation cd88b479-ee21-454e-8851-c3ec323e2127 · inbound

Sign-Aware Gated Sparse Autoencoders: Modeling Anticorrelated Features with Bi-Jump-ReLU Activations cites this paper.

Sign-Aware Gated Sparse Autoencoders: Modeling Anticorrelated Features with Bi-Jump-ReLU Activations Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-04T05:02:51.330615Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 69602583-7167-40d1-924a-e5a348149c10 · inbound

Size Doesn't Matter: Cosine-Scored Sparse Autoencoders cites this paper.

Size Doesn't Matter: Cosine-Scored Sparse Autoencoders Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 27

Resolution: metadata mismatch; arxiv_id, observed 2026-07-01T07:15:29.633755Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation efad8060-df6f-4968-a4f5-5f023a01cae4 · inbound

Critical Percolation as a Synthetic Data Model for Interpretability cites this paper.

Critical Percolation as a Synthetic Data Model for Interpretability Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 30

Resolution: metadata mismatch; arxiv_id, observed 2026-07-04T03:49:29.575560Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation cd273a25-47d3-499a-9d18-b24ccc51c51e · inbound

Do Sparse Autoencoders Learn Meaningful Concept Hierarchies? cites this paper.

Do Sparse Autoencoders Learn Meaningful Concept Hierarchies? Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 24

Resolution: metadata mismatch; arxiv_id, observed 2026-07-04T10:29:45.399426Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation d75a73d9-4b33-479c-9ca4-50e195db219f · inbound

At the Edge of Understanding: Sparse Autoencoders Trace The Limits of Transformer Generalization cites this paper.

At the Edge of Understanding: Sparse Autoencoders Trace The Limits of Transformer Generalization Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 48

Resolution: verified exact; arxiv_id, observed 2026-06-26T01:28:50.567880Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 51ddf6ad-60a4-4f2e-8637-86b23e9d3cc2 · inbound

Surrogate Fidelity: When Can Open LLMs Explain Closed Ones? cites this paper.

Surrogate Fidelity: When Can Open LLMs Explain Closed Ones? Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 41

Resolution: metadata mismatch; arxiv_id, observed 2026-07-01T09:45:40.082546Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-09T06:31:02.800959+00:00.

Observation 322759d2-a828-403a-89b9-196fa3ae6ba1 · inbound

Verbalizable Representations Form a Global Workspace in Language Models cites this paper.

Verbalizable Representations Form a Global Workspace in Language Models Sparse Autoencoders Do Not Find Canonical Units of Analysis

Reference 101

Resolution: unresolved; no resolver link, observed 2026-08-01T23:15:28.686940Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.