Paper Citation Record · LEDGER

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation

As of 13 August 2026, this Paper Citation Record lists 100 of 199 outbound references and 4 inbound Pith citation observations for arXiv:2604.10098.

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

pith.paper-citation-record.v1
2604.10098 v1

Coverage vector

measured 100 of 199 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-05-10T16:17:09.834609Z

measured 104 of 104 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-12T06:34:41.77262+00:00

measured 4 of 4 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-05-20T19:08:23.052621Z

measured 0 of 1 external citation measurements

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

Source: pith, observed 2026-05-20T19:08:54.035553Z

Reference resolution

100 of 199 outbound references displayed

verified exact39
verified fuzzy60
unresolved0
parse uncertain0
malformed identifier0
metadata mismatch1

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation aa24d1ed-4df9-4616-87d0-923a5702cda2 · outbound

This paper cites Attentionisallyouneed.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Attentionisallyouneed

Reference 1

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.177831Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation ee950246-922b-4ce5-b1d6-72867d807dbf · outbound

This paper cites A survey of transformers.AI open, 3:111–132.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation A survey of transformers.AI open, 3:111–132

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.174727Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation cac07e63-6615-443a-843c-fcecb91c42ea · outbound

This paper cites A Survey of Large Language Models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation A Survey of Large Language Models

Reference 3

Resolution: verified exact; local_arxiv, observed 2026-05-11T09:05:57.971328Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation e55857f5-80ff-4e3d-b1c2-7de3f66bbb0d · outbound

This paper cites A survey on vision transformer.IEEE transactions on pattern analysis and machine intelligence, 45(1):87–110.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation A survey on vision transformer.IEEE transactions on pattern analysis and machine intelligence, 45(1):87–110

Reference 4

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.376498Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation c6df39ab-eede-4575-9d58-4364ab95c7f9 · outbound

This paper cites A survey on multimodal large language models.National Science Review, 11(12):nwae403.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation A survey on multimodal large language models.National Science Review, 11(12):nwae403

Reference 5

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.233501Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation cbfe3669-25c5-4ee4-8745-06a0566685d1 · outbound

This paper cites A survey of embodied ai: From simulators to research tasks.IEEE Transactions on Emerging Topics in Computational Intelligence, 6(2):230–244.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation A survey of embodied ai: From simulators to research tasks.IEEE Transactions on Emerging Topics in Computational Intelligence, 6(2):230–244

Reference 6

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.195014Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation b88c1003-3aee-431e-9f69-9ec7cf38f5ed · outbound

This paper cites Longcat-flash technical report.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Longcat-flash technical report

Reference 7

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.303917Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 4d2bcc5e-c999-409d-a858-248bc09478c9 · outbound

This paper cites Longcat-flash-omni technical report.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Longcat-flash-omni technical report

Reference 8

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.048016Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation e76e36bd-6df1-439d-b30f-ab8df2a5736b · outbound

This paper cites Introducing LongCat-flash-thinking: A technical report.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Introducing LongCat-flash-thinking: A technical report

Reference 9

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.275862Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 4020cc53-582e-4bc0-bb4a-2caf4c0fa977 · outbound

This paper cites Longcat-flash-thinking-2601 technical report.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Longcat-flash-thinking-2601 technical report

Reference 10

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.153487Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation dcf4fd8c-f2b9-4198-bd7b-0bf4200a0326 · outbound

This paper cites Vggt: Visual geometry grounded transformer.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Vggt: Visual geometry grounded transformer

Reference 11

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.167958Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation d8e7aaf8-2c98-42fe-9b6b-924dbb82e4e3 · outbound

This paper cites Xstreamvggt: Extremely memory-efficient streaming vision geometry grounded transformer with kv cache compression.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Xstreamvggt: Extremely memory-efficient streaming vision geometry grounded transformer with kv cache compression

Reference 12

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.055150Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 32819257-1579-4fa8-89d7-f924e2c465ef · outbound

This paper cites MapAnything: Universal Feed-Forward Metric 3D Reconstruction.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation MapAnything: Universal Feed-Forward Metric 3D Reconstruction

Reference 13

Resolution: verified exact; arxiv_id, observed 2026-05-12T11:06:15.374425Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 32f393c0-5e39-4dad-b06e-12f81a13b2eb · outbound

This paper cites A survey on model compression for large language models.Transactions of the Association for Computational Linguistics, 12:1556–1577.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation A survey on model compression for large language models.Transactions of the Association for Computational Linguistics, 12:1556–1577

Reference 14

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.181587Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 168935be-15c8-480e-bfcb-529ffdd705f6 · outbound

This paper cites Efficient large language models: A survey.Transactions on Machine Learning Research.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Efficient large language models: A survey.Transactions on Machine Learning Research

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.191707Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation d84989f8-3c51-41bb-9b67-bf363b2c12e7 · outbound

This paper cites Kvquant: Towards 10 million context length llm inference with kv cache quantization.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Kvquant: Towards 10 million context length llm inference with kv cache quantization

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.478668Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation b8481f70-56b4-4759-a618-a19ecdabd51a · outbound

This paper cites Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models

Reference 17

Resolution: verified exact; local_arxiv, observed 2026-05-11T09:05:58.188305Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation aa58d049-a37f-4072-9b53-63927d93ef9c · outbound

This paper cites Efficient attention mechanisms for large language models: A survey.Visual Intelligence.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Efficient attention mechanisms for large language models: A survey.Visual Intelligence

Reference 18

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.576481Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation e44bbf8d-916b-4222-9cb8-1119b1d06ad6 · outbound

This paper cites Speed Always Wins: A Survey on Efficient Architectures for Large Language Models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Speed Always Wins: A Survey on Efficient Architectures for Large Language Models

Reference 19

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.088861Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation f81ad7ad-7f82-475f-93e8-0f3a5f87a551 · outbound

This paper cites Gated delta networks: Improving mamba2 with delta rule.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Gated delta networks: Improving mamba2 with delta rule

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.360005Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation de6dd072-f4d6-4212-b67c-d5e29b8340ac · outbound

This paper cites Kimi Linear: An Expressive, Efficient Attention Architecture.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Kimi Linear: An Expressive, Efficient Attention Architecture

Reference 21

Resolution: verified exact; arxiv_id, observed 2026-05-13T23:49:11.451207Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 39064cea-29aa-4a0c-916c-d3c497d5bbcb · outbound

This paper cites Titans: Learning to memorize at test time.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Titans: Learning to memorize at test time

Reference 22

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.335021Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 1f507c2f-d205-4357-bc1c-f7dd9b5705f1 · outbound

This paper cites SnapMLA: Efficient Long-Context MLA Decoding via Hardware-Aware FP8 Quantized Pipelining.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation SnapMLA: Efficient Long-Context MLA Decoding via Hardware-Aware FP8 Quantized Pipelining

Reference 23

Resolution: verified exact; local_arxiv, observed 2026-05-11T09:05:58.161185Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 393f78f2-11e0-41ad-bb1d-5b6641c983c1 · outbound

This paper cites Efficientstreaminglanguage models with attention sinks.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Efficientstreaminglanguage models with attention sinks

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.241116Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 1efefa74-c70c-405c-ad0e-5de673143071 · outbound

This paper cites When attention sink emerges in language models: An empirical view.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation When attention sink emerges in language models: An empirical view

Reference 25

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.257031Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 6760d8d1-b4b8-4e22-83cd-d258739857b9 · outbound

This paper cites Gated attention for large language models: Non-linearity, sparsity, and attention-sink-free.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Gated attention for large language models: Non-linearity, sparsity, and attention-sink-free

Reference 26

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.252877Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation db85d0cf-59e9-4f09-946c-4e1f39b1bb09 · outbound

This paper cites Why do llms attend to the first token? InSecond Conference on Language Modeling.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Why do llms attend to the first token? InSecond Conference on Language Modeling

Reference 27

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.214422Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation a5a07e3f-2cfc-4b2c-8199-fe04345f26d2 · outbound

This paper cites Kvsink: Understanding and enhancing the preservation of attention sinks in kv cache quantization for llms.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Kvsink: Understanding and enhancing the preservation of attention sinks in kv cache quantization for llms

Reference 28

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.299064Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 3df883b1-4497-4f1a-bca7-dfb63386f7d5 · outbound

This paper cites Quantizable transformers: Removing outliers by helping attention heads do nothing.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Quantizable transformers: Removing outliers by helping attention heads do nothing

Reference 29

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.396459Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation a50651e7-c8f5-4354-8398-d541d342ac69 · outbound

This paper cites Don't Deceive Me: Mitigating Gaslighting through Attention Reallocation in LMMs.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Don't Deceive Me: Mitigating Gaslighting through Attention Reallocation in LMMs

Reference 30

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.172892Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 65866687-b36b-4dd0-8b12-8914262073ad · outbound

This paper cites Attention reallocation: Towards zero-cost and controllable hallucination mitigation of mllms.International Journal of Computer Vision, 134(1):22.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Attention reallocation: Towards zero-cost and controllable hallucination mitigation of mllms.International Journal of Computer Vision, 134(1):22

Reference 31

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.443368Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation bfd8a969-39f3-4a61-a07e-c33628358d1a · outbound

This paper cites Vasparse: Towards efficient visual hallucination mitigation via visual-aware token sparsification.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Vasparse: Towards efficient visual hallucination mitigation via visual-aware token sparsification

Reference 32

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.568874Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation e2987534-74e4-409c-8891-857451b67892 · outbound

This paper cites Forget- ting to forget: Attention sink as a gateway for backdoor- ing llm unlearning.arXiv preprint arXiv:2510.17021.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Forget- ting to forget: Attention sink as a gateway for backdoor- ing llm unlearning.arXiv preprint arXiv:2510.17021

Reference 33

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.194894Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation f6a1d808-9c04-4d7d-87d2-01eb8a530128 · outbound

This paper cites Interpreting the repeated token phenomenon in large language models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Interpreting the repeated token phenomenon in large language models

Reference 34

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.338513Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation e045dc42-cd84-4e90-b98e-5196a0f5e5ec · outbound

This paper cites Leveraging registers in vision trans- formers for robust adaptation.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Leveraging registers in vision trans- formers for robust adaptation

Reference 35

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.218447Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation f834c925-b530-43f9-9a58-04361dd44fce · outbound

This paper cites Nosa: Native and offloadable sparse attention.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Nosa: Native and offloadable sparse attention

Reference 36

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.272183Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation e446f852-d717-4982-bde7-a17c29e5d233 · outbound

This paper cites OBCache: Optimal Brain KV Cache Pruning for Efficient Long-Context LLM Inference.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation OBCache: Optimal Brain KV Cache Pruning for Efficient Long-Context LLM Inference

Reference 37

Resolution: verified exact; arxiv_id, observed 2026-06-01T02:02:20.220338Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 9f4a8eca-af8a-46d6-a13f-a952e0fd9773 · outbound

This paper cites SALS: Sparse attention in latent space for KV cache compression.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation SALS: Sparse attention in latent space for KV cache compression

Reference 38

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.341908Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 2137741a-cb04-4054-9a6c-7edd591f158b · outbound

This paper cites OjaKV: Context-Aware Online Low-Rank KV Cache Compression.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation OjaKV: Context-Aware Online Low-Rank KV Cache Compression

Reference 39

Resolution: verified exact; local_arxiv, observed 2026-05-11T09:05:57.862495Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation f93cbd5e-34d1-4dee-9543-fbd0992eb31e · outbound

This paper cites Mitigating attention sinks and massive activations in audio-visual speech recognition with llms.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Mitigating attention sinks and massive activations in audio-visual speech recognition with llms

Reference 40

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.141700Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 771e016f-74eb-4e44-8c0f-fd8cd0748642 · outbound

This paper cites Attention sinks and compression valleys in llms are two sides of the same coin.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Attention sinks and compression valleys in llms are two sides of the same coin

Reference 41

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.486196Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 98a8b51d-41ea-4fb7-972b-0e6d7e477954 · outbound

This paper cites Outlier-safe pre-training for robust 4-bit quantization of large language models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Outlier-safe pre-training for robust 4-bit quantization of large language models

Reference 42

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.496430Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 768890e9-e056-4db0-bb79-bcb2b7f7d8b9 · outbound

This paper cites Unveiling super experts in mixture-of-experts large language models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Unveiling super experts in mixture-of-experts large language models

Reference 43

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.506535Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation ef20d4e6-ce66-462f-ae9d-c9befa908093 · outbound

This paper cites Value-state gated attention for mitigating extreme-token phenomena in transformers.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Value-state gated attention for mitigating extreme-token phenomena in transformers

Reference 44

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.839886Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 71fe4d71-65e8-45b8-a10f-4c853a3c44d1 · outbound

This paper cites Research and latest advancements.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Research and latest advancements

Reference 45

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.200975Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation d68d2ff5-7e28-4169-a9ba-e57d6582fa6b · outbound

This paper cites What are you sinking? a geometric approach on attention sink.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation What are you sinking? a geometric approach on attention sink

Reference 46

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.554852Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 114ccecc-d602-430e-ac76-ba5dc9bf27a2 · outbound

This paper cites CTR-Sink: Attention Sink for Language Models in Click-Through Rate Prediction.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation CTR-Sink: Attention Sink for Language Models in Click-Through Rate Prediction

Reference 47

Resolution: verified exact; arxiv_id, observed 2026-06-03T02:05:41.714027Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 1bdedb93-1215-4d49-b290-1109a8048d31 · outbound

This paper cites Does roberta perform better than bert in continual learning: An attention sink perspective.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Does roberta perform better than bert in continual learning: An attention sink perspective

Reference 48

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.409523Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation ea2d9ab0-7a6b-4407-a97e-11bea1f4a85a · outbound

This paper cites Outlier dimensions that disrupttransformersaredrivenbyfrequency.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Outlier dimensions that disrupttransformersaredrivenbyfrequency

Reference 49

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.447264Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 2f179ad1-0e5e-4064-96d6-24c8ead91245 · outbound

This paper cites Understanding and overcoming the challenges of efficient transformer quantization.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Understanding and overcoming the challenges of efficient transformer quantization

Reference 50

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.380628Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 246e340d-2548-400b-8953-3ecb2b5c7c8c · outbound

This paper cites Bert busters: Outlier dimensions that disrupt transformers.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Bert busters: Outlier dimensions that disrupt transformers

Reference 51

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.413020Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 5a3fc787-8962-4e36-b383-f53c3f750ef3 · outbound

This paper cites Positional artefacts propagate through masked language model embeddings.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Positional artefacts propagate through masked language model embeddings

Reference 52

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.557840Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation cf02788f-8034-429b-b9f5-4aa2a28ca8d7 · outbound

This paper cites What does bert look at? an analysis of bert’s attention.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation What does bert look at? an analysis of bert’s attention

Reference 53

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.403131Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 960f5649-5ff8-4d3e-8efa-a60693af1ec3 · outbound

This paper cites MiMo-V2-Flash Technical Report.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation MiMo-V2-Flash Technical Report

Reference 54

Resolution: verified exact; arxiv_id, observed 2026-05-12T11:33:32.947936Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 2db3180d-4542-4554-bdf8-9c6aea642ede · outbound

This paper cites Attention needs to focus: A unified perspective on attention allocation.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Attention needs to focus: A unified perspective on attention allocation

Reference 55

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.925228Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation be95aad9-6650-4018-be11-c998d820fb7e · outbound

This paper cites On the existence and behaviour of secondary attention sinks.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation On the existence and behaviour of secondary attention sinks

Reference 56

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.834936Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 1854e6bb-ec19-4971-b475-4d91bd360291 · outbound

This paper cites Sliding window attention adaptation.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Sliding window attention adaptation

Reference 57

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.899852Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation f3c2dcbf-fbba-4479-8225-4d5f44fb2955 · outbound

This paper cites Dope: Denoising rotary position embedding.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Dope: Denoising rotary position embedding

Reference 58

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.932730Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation f37dca4a-b180-431c-8a22-8f35b8cf15ee · outbound

This paper cites Tweo: Transformers without extreme outliers enables fp8 training and quantization for dummies.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Tweo: Transformers without extreme outliers enables fp8 training and quantization for dummies

Reference 59

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.888361Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation d31aa59b-3ce7-434b-a2db-31062d808ae7 · outbound

This paper cites Lost in the middle: An emergent property from information retrieval demands in LLMs.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Lost in the middle: An emergent property from information retrieval demands in LLMs

Reference 60

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.983080Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 3f1037da-5a52-4862-bdac-73ab5e75d78f · outbound

This paper cites Hybrid Architectures for Language Models: Systematic Analysis and Design Insights.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Hybrid Architectures for Language Models: Systematic Analysis and Design Insights

Reference 61

Resolution: verified exact; local_arxiv, observed 2026-05-11T09:05:58.030354Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 50321c96-3347-433c-9926-0d01d8d28f19 · outbound

This paper cites CacheClip: Accelerating RAG with Effective KV Cache Reuse.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation CacheClip: Accelerating RAG with Effective KV Cache Reuse

Reference 62

Resolution: verified exact; arxiv_id, observed 2026-05-22T02:03:50.604177Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 25c2da19-77a3-45dc-b3c5-5851a371073f · outbound

This paper cites Artificial hippocampus networks for efficient long-context modeling.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Artificial hippocampus networks for efficient long-context modeling

Reference 63

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.940007Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 207ef988-1d6f-4bac-86f3-3fd01dc26dba · outbound

This paper cites vattention: Verified sparse attention via sampling.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation vattention: Verified sparse attention via sampling

Reference 64

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.237214Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 54bf7f7a-fb52-4402-b64b-bc7c801bcdac · outbound

This paper cites All for one: Llms solve mental math at the last token with information transferred from other tokens.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation All for one: Llms solve mental math at the last token with information transferred from other tokens

Reference 65

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.269537Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 03d321b5-84f7-41f9-898f-29e2f3a152f9 · outbound

This paper cites gpt-oss-120b & gpt-oss-20b Model Card.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation gpt-oss-120b & gpt-oss-20b Model Card

Reference 66

Resolution: verified exact; local_arxiv, observed 2026-05-11T09:05:58.295555Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation ff7cee7d-36e9-4391-b84a-5505e1d91cb2 · outbound

This paper cites Integral transformer: Denoising attention, not too much not too little.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Integral transformer: Denoising attention, not too much not too little

Reference 67

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.302360Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 336dfe7d-52d5-45d8-bbdb-ef3468de23b6 · outbound

This paper cites H2eal: Hybrid-bonding architecture with hybrid sparse attention for efficient long-context llm inference.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation H2eal: Hybrid-bonding architecture with hybrid sparse attention for efficient long-context llm inference

Reference 68

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.164002Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 5a3aa25d-3f69-4882-8886-3b6d2f0c7957 · outbound

This paper cites Accelerating Prefilling via Decoding-time Contribution Sparsity.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Accelerating Prefilling via Decoding-time Contribution Sparsity

Reference 69

Resolution: verified exact; local_arxiv, observed 2026-05-11T09:05:58.042049Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 9ec1466d-6bf2-48ed-932c-5caa6f95b8d0 · outbound

This paper cites Orthorank: Token selection via sink token orthogonality for efficient llm inference.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Orthorank: Token selection via sink token orthogonality for efficient llm inference

Reference 70

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.320335Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation e2bde12f-8c3e-4e22-9bb3-ff06e7bdbfcd · outbound

This paper cites Earn: Efficient inference acceleration for llm-based generative recommendation by register tokens.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Earn: Efficient inference acceleration for llm-based generative recommendation by register tokens

Reference 71

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.349150Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation b9c372f8-f66e-4582-b2d8-0b483a5920f6 · outbound

This paper cites DeltaLLM: A Training-Free Framework Exploiting Temporal Sparsity for Efficient Edge LLM Inference.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation DeltaLLM: A Training-Free Framework Exploiting Temporal Sparsity for Efficient Edge LLM Inference

Reference 72

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.881229Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 795242bf-3d0f-49a8-96ad-9f827583f903 · outbound

This paper cites Two Heads Are Better than One: Simulating Large Transformers with Small Ones.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Two Heads Are Better than One: Simulating Large Transformers with Small Ones

Reference 73

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.232597Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation adc68189-911a-4ea2-bb71-b3f964fbeaf9 · outbound

This paper cites Learn from the Past: Fast Sparse Indexing for Large Language Model Decoding.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Learn from the Past: Fast Sparse Indexing for Large Language Model Decoding

Reference 74

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.988048Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 1cf09c47-39bd-4361-80bd-044d0ef29ddf · outbound

This paper cites Zerotuning: Unlocking the initial token’s power to enhance large language models without training.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Zerotuning: Unlocking the initial token’s power to enhance large language models without training

Reference 75

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.561092Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 1095e0a4-1e1c-49cf-a782-17eca38f0219 · outbound

This paper cites Delta attention: Fast and accurate sparse attention inference by delta correction.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Delta attention: Fast and accurate sparse attention inference by delta correction

Reference 76

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.009864Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 1639b654-9aab-48e4-bd4f-b3999b366418 · outbound

This paper cites Softpick: No Attention Sink, No Massive Activations with Rectified Softmax.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Softpick: No Attention Sink, No Massive Activations with Rectified Softmax

Reference 77

Resolution: verified exact; local_arxiv, observed 2026-05-11T09:05:58.105206Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 3ee72983-7d7e-4f35-8819-1d5a02246a15 · outbound

This paper cites Keydiff: Key similarity-based kv cache eviction for long-context llm inference in resource-constrained environments.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Keydiff: Key similarity-based kv cache eviction for long-context llm inference in resource-constrained environments

Reference 78

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.591502Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation b2b788cb-8714-4281-9d76-4ce1c7ff662d · outbound

This paper cites Edgeinfinite: A memory-efficient infinite-context transformer for edge devices.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Edgeinfinite: A memory-efficient infinite-context transformer for edge devices

Reference 79

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.539087Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation da0721a4-2131-4411-a749-dce0a4205400 · outbound

This paper cites Efficient many-shot in-context learning with dynamic block-sparse attention.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Efficient many-shot in-context learning with dynamic block-sparse attention

Reference 80

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.516167Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 84ca17e2-b00a-40d1-adc7-abaa8a0af1db · outbound

This paper cites On the emergence of position bias in transformers.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation On the emergence of position bias in transformers

Reference 81

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.529286Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 7f68eefa-308c-4eee-8c79-407d050c2206 · outbound

This paper cites Systematic outliers in large language models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Systematic outliers in large language models

Reference 82

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.541810Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 504b86fc-5c12-47e3-a090-8a0a6792d74f · outbound

This paper cites Sliding Window Attention Training for Efficient Large Language Models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Sliding Window Attention Training for Efficient Large Language Models

Reference 83

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.944778Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 7897ecfb-2b3e-4b52-bc2a-5b13b7ea0df4 · outbound

This paper cites Rotatekv: Accurate and robust 2-bit kv cache quantization for llms via outlier-aware adaptive rotations.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Rotatekv: Accurate and robust 2-bit kv cache quantization for llms via outlier-aware adaptive rotations

Reference 84

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.588022Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation fb8c00e9-eaeb-46e1-915f-b24ca2ef0ae7 · outbound

This paper cites Weight-based analysis of detokenization in language models: Understanding the first stage of inference without inference.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Weight-based analysis of detokenization in language models: Understanding the first stage of inference without inference

Reference 85

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.583256Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 84eace44-4527-4779-9b34-87224001a7b8 · outbound

This paper cites Task-KV: Task-aware KV Cache Optimization via Semantic Differentiation of Attention Heads.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Task-KV: Task-aware KV Cache Optimization via Semantic Differentiation of Attention Heads

Reference 86

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:58.130976Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 603475af-914f-474d-a411-ae9f945955aa · outbound

This paper cites LLMs Know What to Drop: Self-Attention Guided KV Cache Eviction for Efficient Long-Context Inference.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation LLMs Know What to Drop: Self-Attention Guided KV Cache Eviction for Efficient Long-Context Inference

Reference 87

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T09:05:58.252809Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 91ddf077-214f-423d-ad7b-374bb4a8388d · outbound

This paper cites Unigist: Towards general and hardware-aligned sequence-level long context compression.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Unigist: Towards general and hardware-aligned sequence-level long context compression

Reference 88

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.489633Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 27c70ef9-2217-497b-ad11-3b63e1763b40 · outbound

This paper cites Cache me if you must: Adaptive key-value quantization for large language models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Cache me if you must: Adaptive key-value quantization for large language models

Reference 89

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.532382Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 89d228a3-631e-47a7-842f-9e04d8bc6c14 · outbound

This paper cites Position bias mitigates position bias: Mitigate position bias through inter-position knowledge distillation.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Position bias mitigates position bias: Mitigate position bias through inter-position knowledge distillation

Reference 90

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.466482Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 3bbb1062-47a9-45c6-8b6c-5a19fd9ed013 · outbound

This paper cites Attention sinks: A’catch, tag, release’mechanism for embeddings.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Attention sinks: A’catch, tag, release’mechanism for embeddings

Reference 91

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.352899Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 0ccd0280-dc05-4452-90db-33226c650173 · outbound

This paper cites The singular anchor: First token dominance in large language model attention sinks.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation The singular anchor: First token dominance in large language model attention sinks

Reference 92

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.264328Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 16d92e27-cecf-43db-ac4e-ec692a42a7c8 · outbound

This paper cites Attention entropy is a key factor: An analysis of parallel context encoding with full-attention-based pre-trained language models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Attention entropy is a key factor: An analysis of parallel context encoding with full-attention-based pre-trained language models

Reference 93

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.287632Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 91e055e8-98f2-4bc1-b6cf-dbd9e1794167 · outbound

This paper cites Sgd-kv: Summarization guided kv cache compression.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Sgd-kv: Summarization guided kv cache compression

Reference 94

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.597540Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation e16d1cbe-8c4b-4ce2-ae7b-270cd756d960 · outbound

This paper cites Initial-key cache: An efficient kv cache strategy focusing on initial and key tokens for llms.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Initial-key cache: An efficient kv cache strategy focusing on initial and key tokens for llms

Reference 95

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.482687Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation e660d061-3644-48d4-9a6c-d5993d303167 · outbound

This paper cites Entropy-guided kv caching for efficient llm inference.Mathematics, 13(15):2366.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Entropy-guided kv caching for efficient llm inference.Mathematics, 13(15):2366

Reference 96

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.434982Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation c5a3008c-c18e-4efb-b018-f2ed58490db3 · outbound

This paper cites Subkv: Quantizing long context kv cache for sub-billion parameter language models on edge devices.Software: Practice and Experience.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Subkv: Quantizing long context kv cache for sub-billion parameter language models on edge devices.Software: Practice and Experience

Reference 97

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.535584Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 4e853cd0-58bf-4c20-9f95-edc29fb87b1e · outbound

This paper cites Massive activations in large language models.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Massive activations in large language models

Reference 98

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.503056Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation f6c75b00-d2a4-49f4-995c-699f964361e9 · outbound

This paper cites Omnisparse: Training-aware fine-grained sparse attention for long-video mllms.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Omnisparse: Training-aware fine-grained sparse attention for long-video mllms

Reference 99

Resolution: verified exact; arxiv_id, observed 2026-05-11T09:05:57.996556Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 8268f3fa-58e1-4791-b138-81a3cabacf43 · outbound

This paper cites Bitmar: Low-bit multimodal fusion with episodic memory for edge devices.

Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation Bitmar: Low-bit multimodal fusion with episodic memory for edge devices

Reference 100

Resolution: verified fuzzy; raw_fallback, observed 2026-05-17T16:04:57.331581Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Pith citing papers

Observation 4fdc6361-eb0e-4b00-a440-7d96b8680378 · inbound

When Does Value-Aware KV Eviction Help? A Fixed-Contract Diagnostic for Non-Monotone Cache Compression cites this paper.

When Does Value-Aware KV Eviction Help? A Fixed-Contract Diagnostic for Non-Monotone Cache Compression Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation

Reference 42

Resolution: verified exact; local_arxiv, observed 2026-05-12T07:51:42.864258Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 8922d645-bb58-49af-94e9-c1af7a027842 · inbound

Priming: Hybrid State Space Models From Pre-trained Transformers cites this paper.

Priming: Hybrid State Space Models From Pre-trained Transformers Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation

Reference 81

Resolution: verified exact; local_arxiv, observed 2026-05-12T08:21:24.357221Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 4f600ed0-32bd-4188-b9ed-d20e479b24bd · inbound

Registers Matter for Pixel-Space Diffusion Transformers cites this paper.

Registers Matter for Pixel-Space Diffusion Transformers Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation

Reference 29

Resolution: verified exact; local_arxiv, observed 2026-05-20T19:08:54.036916Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.

Observation 4c0a1a53-3f16-4983-80fb-9df02a31f889 · inbound

OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond cites this paper.

OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation

Reference 50

Resolution: verified exact; local_arxiv, observed 2026-05-20T07:58:07.543129Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-12T06:34:41.77262+00:00.