Paper Citation Record · LEDGER

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models

As of 20 August 2026, this Paper Citation Record lists 35 of 35 outbound references and 1 inbound Pith citation observation for arXiv:2412.02237.

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

pith.paper-citation-record.v1
2412.02237 v3

Coverage vector

measured 35 of 35 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-11T23:49:50.312022Z

measured 36 of 36 standing notices

One-hop event checks from named stored sources.

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

measured 1 of 1 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-05-10T19:16:51.134397Z

measured 0 of 1 external citation measurements

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

Source: arxiv_reference, observed 2026-05-10T23:15:47.690569Z

Reference resolution

35 of 35 outbound references displayed

verified exact0
verified fuzzy14
unresolved20
parse uncertain0
malformed identifier1
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 92c095c2-2ebe-448f-960f-96e07159766c · outbound

This paper cites Table 11 lists the 12 animals and 10 colors used to generate these prompts, with the full prompt list available in our core codebase.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Table 11 lists the 12 animals and 10 colors used to generate these prompts, with the full prompt list available in our core codebase

Reference 1

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.511328Z

Source-reported events for the cited work

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

Observation 24a30692-1a84-44b5-b25c-c8fc1418addb · outbound

This paper cites an unresolved cited work.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Unresolved cited work

Reference 3

Resolution: unresolved; raw_fallback, observed 2026-08-11T23:49:50.657393Z

Source-reported events for the cited work

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

Observation a449c785-1108-49e3-b2f7-b1e0e1157bad · outbound

This paper cites A larger (LeRHF − MoRHF) area indicates that the ordering of CA heads better reflects the relevance of the corresponding concept.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models A larger (LeRHF − MoRHF) area indicates that the ordering of CA heads better reflects the relevance of the corresponding concept

Reference 4

Resolution: malformed identifier; raw_fallback, observed 2026-08-11T23:49:50.632683Z

Source-reported events for the cited work

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

Observation 2affa6b2-1337-40e0-9b4c-c166b7e612bf · outbound

This paper cites Prompt-to-Prompt Image Editing with Cross Attention Control.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Prompt-to-Prompt Image Editing with Cross Attention Control

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.190975Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ed5c1671-80ef-423c-a4c7-0d865bf88030 · outbound

This paper cites SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equations.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equations

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.199522Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9516ee1c-2995-4d6b-949a-c7860ee30310 · outbound

This paper cites Count how many of the following five images contain{intended meaning of the polysemous word} but no {unintended meaning of the polysemous word }.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Count how many of the following five images contain{intended meaning of the polysemous word} but no {unintended meaning of the polysemous word }

Reference 8

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.608873Z

Source-reported events for the cited work

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

Observation c611356c-e6bb-4176-9519-6c9bc2df152f · outbound

This paper cites DINOv2: Learning Robust Visual Features without Supervision.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models DINOv2: Learning Robust Visual Features without Supervision

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.207748Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 65e66eb9-74a5-4315-bc62-9baeedbda4cc · outbound

This paper cites Zero-shot image-to-image translation.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Zero-shot image-to-image translation

Reference 10

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.832220Z

Source-reported events for the cited work

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

Observation f54045b0-7e26-4043-98b9-6cce7dd88309 · outbound

This paper cites Controlling generative models with continuous factors of variations.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Controlling generative models with continuous factors of variations

Reference 11

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.819854Z

Source-reported events for the cited work

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

Observation aad47d13-616c-4a73-8691-89701026f704 · outbound

This paper cites PromptHero.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models PromptHero

Reference 12

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.807190Z

Source-reported events for the cited work

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

Observation d772e344-90f0-4b8b-bb53-3b4d47206d8d · outbound

This paper cites SAM 2: Segment Anything in Images and Videos.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models SAM 2: Segment Anything in Images and Videos

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.223823Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c58dc8cd-0d7c-4886-8194-d2d5d2de86a8 · outbound

This paper cites Grounded SAM: Assembling Open-World Models for Diverse Visual Tasks.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Grounded SAM: Assembling Open-World Models for Diverse Visual Tasks

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.228272Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d9716935-c1f1-4055-8c6d-686cfc7f74f3 · outbound

This paper cites U-net: Convolutional networks for biomed- ical image segmentation.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models U-net: Convolutional networks for biomed- ical image segmentation

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.232080Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation aee6471b-8488-4f5d-9d06-8a041e7d822a · outbound

This paper cites 13 Published as a conference paper at ICLR 2025 Richard Tomsett, Dan Harborne, Supriyo Chakraborty, Prudhvi Gurram, and Alun Preece.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models 13 Published as a conference paper at ICLR 2025 Richard Tomsett, Dan Harborne, Supriyo Chakraborty, Prudhvi Gurram, and Alun Preece

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.786614Z

Source-reported events for the cited work

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

Observation b4234a20-6d6a-4b9b-ab8b-8d302df4e979 · outbound

This paper cites URL https://ojs.aaai.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models URL https://ojs.aaai

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.239735Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c904da17-0cac-4b30-af6a-e1b34458bed5 · outbound

This paper cites 17 B.2 Role of the argmax operation in HRV construction.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models 17 B.2 Role of the argmax operation in HRV construction

Reference 18

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.669282Z

Source-reported events for the cited work

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

Observation 4e65042b-a6d6-41d2-b742-df6399353368 · outbound

This paper cites an unresolved cited work.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Unresolved cited work

Reference 19

Resolution: unresolved; raw_fallback, observed 2026-08-11T23:49:50.620830Z

Source-reported events for the cited work

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

Observation 57d8313a-7d23-441b-9904-bde3b239496b · outbound

This paper cites The MoRHF weakening rapidly removes concept-relevant content, whereas the LeRHF weakening either preserves the original image longer or removes irrelevant content first.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models The MoRHF weakening rapidly removes concept-relevant content, whereas the LeRHF weakening either preserves the original image longer or removes irrelevant content first

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.645382Z

Source-reported events for the cited work

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

Observation ced8a51d-0000-48d6-acef-0ecbf6bdf1a9 · outbound

This paper cites Both methods are evaluated on the Color editing benchmark with SA replacement rates varying from 0.0 to 1.0.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Both methods are evaluated on the Color editing benchmark with SA replacement rates varying from 0.0 to 1.0

Reference 23

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.549896Z

Source-reported events for the cited work

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

Observation 977d496b-e868-4c37-b26a-88da8ebb22fd · outbound

This paper cites an unresolved cited work.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Unresolved cited work

Reference 24

Resolution: unresolved; raw_fallback, observed 2026-08-11T23:49:50.597066Z

Source-reported events for the cited work

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

Observation 2a0b06d5-5e65-4ace-9db6-cd0ca623baaa · outbound

This paper cites The CLIP score measures the CLIP image-text similarity between the edited image and the target prompt, assessing how well the edited image aligns with the target prompt.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models The CLIP score measures the CLIP image-text similarity between the edited image and the target prompt, assessing how well the edited image aligns with the target prompt

Reference 25

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.585509Z

Source-reported events for the cited work

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

Observation bec48f22-2c88-4a3a-bbda-68bb9609e62e · outbound

This paper cites While prior works (Parmar et al., 2023; Kim et al.,.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models While prior works (Parmar et al., 2023; Kim et al.,

Reference 26

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.573526Z

Source-reported events for the cited work

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

Observation 3ffabbd8-7881-47f2-9a67-67ffa264a760 · outbound

This paper cites an unresolved cited work.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Unresolved cited work

Reference 27

Resolution: unresolved; raw_fallback, observed 2026-08-11T23:49:50.561425Z

Source-reported events for the cited work

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

Observation 38d2f98f-97a8-46c8-954a-2b9fdb13affd · outbound

This paper cites 34 Published as a conference paper at ICLR 2025 Figure 23: Qualitative results of image editing comparing P2P (Hertz et al.,.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models 34 Published as a conference paper at ICLR 2025 Figure 23: Qualitative results of image editing comparing P2P (Hertz et al.,

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.537698Z

Source-reported events for the cited work

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

Observation 070311c1-4766-4fe1-a3cf-c5e9126bdade · outbound

This paper cites an unresolved cited work.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Unresolved cited work

Reference 30

Resolution: unresolved; raw_fallback, observed 2026-08-11T23:49:50.525315Z

Source-reported events for the cited work

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

Observation 43bcdd47-7588-4a10-a52d-faff793fce29 · outbound

This paper cites an unresolved cited work.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Unresolved cited work

Reference 32

Resolution: unresolved; raw_fallback, observed 2026-08-11T23:49:50.499108Z

Source-reported events for the cited work

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

Observation 2ee4b9b6-9b5a-4145-a4ae-71b50a5d395f · outbound

This paper cites an unresolved cited work.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Unresolved cited work

Reference 33

Resolution: unresolved; raw_fallback, observed 2026-08-11T23:49:50.486695Z

Source-reported events for the cited work

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

Observation d13fbbce-5081-4796-8e70-3dec7273f723 · outbound

This paper cites We then construct HRVs individually for both the 34-concept and 35-concept sets and compare them through visualization.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models We then construct HRVs individually for both the 34-concept and 35-concept sets and compare them through visualization

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.459648Z

Source-reported events for the cited work

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

Observation 05c8e535-2c5c-4e74-8987-249c0aa2f448 · outbound

This paper cites an unresolved cited work.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Unresolved cited work

Reference 46

Resolution: unresolved; raw_fallback, observed 2026-08-11T23:49:50.472413Z

Source-reported events for the cited work

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

Observation cfed7b2c-ce98-4317-9db4-65850c8da78e · outbound

This paper cites Imagenet: A large-scale hi- erarchical image database.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Imagenet: A large-scale hi- erarchical image database

Reference 2018

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.172280Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6598efbf-63fd-4e84-ae29-b8174c238272 · outbound

This paper cites https://distill.pub/2020/circuits/zoom-in.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models https://distill.pub/2020/circuits/zoom-in

Reference 2020

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.203777Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 87bb6083-fcc9-425d-b42a-d75a154b3ec0 · outbound

This paper cites Training-Free Structured Diffusion Guidance for Compositional Text-to-Image Synthesis.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Training-Free Structured Diffusion Guidance for Compositional Text-to-Image Synthesis

Reference 2021

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.177657Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a7887e27-e5d4-4e47-b710-a23a34539aeb · outbound

This paper cites Interpreting CLIP's Image Representation via Text-Based Decomposition.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Interpreting CLIP's Image Representation via Text-Based Decomposition

Reference 2022

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.182217Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7e2ccdb6-cc52-46dc-8706-731f2356e5ca · outbound

This paper cites Multimodal neurons in artificial neural networks.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Multimodal neurons in artificial neural networks

Reference 2023

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T23:49:50.845753Z

Source-reported events for the cited work

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

Observation 076b8705-0302-40fb-9955-16b675d9d79c · outbound

This paper cites Pseudo Numerical Methods for Diffusion Models on Manifolds.

Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models Pseudo Numerical Methods for Diffusion Models on Manifolds

Reference 2024

Resolution: unresolved; no resolver link, observed 2026-08-11T23:49:50.195319Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation bf81153b-0dfc-4785-baa4-b752cb66e116 · inbound

Selective Aggregation of Attention Maps Improves Diffusion-Based Visual Interpretation cites this paper.

Selective Aggregation of Attention Maps Improves Diffusion-Based Visual Interpretation Cross-Attention Head Position Patterns Can Align with Human Visual Concepts in Text-to-Image Generative Models

Reference 7

Resolution: verified exact; arxiv_id, observed 2026-05-10T23:15:47.693742Z

Source-reported events for the cited work

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