Paper Citation Record · LEDGER

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation

As of 9 August 2026, this Paper Citation Record lists 33 of 33 outbound references and 8 inbound Pith citation observations for arXiv:2508.02401.

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

pith.paper-citation-record.v1
2508.02401 v1

Coverage vector

measured 33 of 33 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-06T05:00:04.960649Z

measured 41 of 41 standing notices

One-hop event checks from named stored sources.

Source: scholarly_work_events, retraction_status_cache, observed 2026-08-08T06:32:00.761636+00:00

measured 8 of 8 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-07T00:45:48.497849Z

measured 0 of 1 external citation measurements

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

Source: arxiv_reference, observed 2026-07-04T16:49:58.010890Z

Reference resolution

33 of 33 outbound references displayed

verified exact0
verified fuzzy2
unresolved31
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

No source-named external measurement is stored.

Outbound references

Observation 3a02f6cf-deb4-4025-aa64-a653838170ff · outbound

This paper cites , " * write output.state after.block = add.period write newline.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation , " * write output.state after.block = add.period write newline

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.779838Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 27f24123-5625-4518-bb6d-ade1b4f74603 · outbound

This paper cites write newline.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation write newline

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.786401Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d35d4ff9-9bcd-4f7c-9ba6-7225b7b00128 · outbound

This paper cites GPT-4 Technical Report.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation GPT-4 Technical Report

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.793764Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 29409de1-d8e0-4a14-9cb3-d8fe2c293499 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 4

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:06.573417Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation a98a3025-0163-4f9f-889c-a1555ac9658a · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 5

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:06.558665Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 184f0cd2-9907-46dc-b54d-70bb15275c2b · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 6

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:06.392314Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 8c878f50-e796-4a0e-bff9-99634a5bdaa3 · outbound

This paper cites PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funneling.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funneling

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.815941Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 84ed15f5-85b1-4afc-a7ef-2aa92570c212 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 8

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:06.210840Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 26dc0eb0-c611-4196-b4b1-2e5f2e029558 · outbound

This paper cites The Llama 3 Herd of Models.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation The Llama 3 Herd of Models

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.828588Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8f3a3bae-6a3c-45ec-b188-bf28b077236a · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 10

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:06.029166Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 40aa3301-c05b-4827-ae76-844e2120e222 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 11

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.870810Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation d078d48f-b356-4f6f-88ad-62ae9762ad7a · outbound

This paper cites Qwen2.5 Technical Report.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Qwen2.5 Technical Report

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.844683Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 6d5705cf-9e95-4c13-baa3-b7bdb6ecc9bd · outbound

This paper cites From CLIP to DINO: Visual Encoders Shout in Multi-modal Large Language Models.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation From CLIP to DINO: Visual Encoders Shout in Multi-modal Large Language Models

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.849995Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1ced2924-ce9e-4a4f-9fc1-de6b2f1b1e86 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 14

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.717892Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 8c0fae0e-b197-4574-816e-8b75130645d1 · outbound

This paper cites GEAR: An Efficient KV Cache Compression Recipe for Near-Lossless Generative Inference of LLM.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation GEAR: An Efficient KV Cache Compression Recipe for Near-Lossless Generative Inference of LLM

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.860070Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d879762a-4373-4993-a9cb-18c01300240e · outbound

This paper cites W.; Hassoun, J.; Keutzer, K.; and Gholami, A.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation W.; Hassoun, J.; Keutzer, K.; and Gholami, A

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T05:00:05.583593Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 7bcaff39-b696-41ee-b68a-2714272fe34c · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 17

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.551971Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 3f2a0ce4-018a-4b16-807b-aa3d3c12c121 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 18

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.518917Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 13c1d72a-dab8-4db1-8802-b93013336e6d · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 19

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.474642Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation e77090e4-efae-4abd-a79c-1ab8e7b9dce8 · outbound

This paper cites In-context Learning and Induction Heads.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation In-context Learning and Induction Heads

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.888968Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation deaabee4-fd31-4276-84ab-77e032f2bc63 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 21

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.433406Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation c15ab8df-5f20-4cc3-b2b8-aa06a8f67add · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 22

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.394304Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 7a4c5a16-b72c-46d0-8de1-a9b4aeae8428 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 23

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.368266Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 8cd0f08a-073b-4a78-908e-d72ddbd2c551 · outbound

This paper cites S.; Mueller, A.; Wallace, B.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation S.; Mueller, A.; Wallace, B

Reference 24

Resolution: verified fuzzy; raw_fallback, observed 2026-08-06T05:00:05.340020Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 1dd5fdba-5518-44ba-b0b2-06ee75489207 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 25

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.312877Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 72e1750b-60d2-4b68-802d-972fcb93b502 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 26

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.286136Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 796bbb22-98d9-44f4-b3b8-d8f223e9624d · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 27

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.267210Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation a1a835be-7f60-47f0-9cdb-b4fcfcaf23e8 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 28

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.249883Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 40396696-54d0-46ab-ac12-01f4f16c3c70 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 29

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.230188Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 02531ed0-ac6e-4216-b7b9-95e72124d791 · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 30

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.211749Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 559f2c11-27d4-4467-990f-fef5903a6315 · outbound

This paper cites Which Attention Heads Matter for In-Context Learning?.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Which Attention Heads Matter for In-Context Learning?

Reference 31

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.950853Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 879b0e84-2314-4f70-8203-4ad5c9b152bf · outbound

This paper cites an unresolved cited work.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Unresolved cited work

Reference 32

Resolution: unresolved; raw_fallback, observed 2026-08-06T05:00:05.192752Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation f850a8c1-c2d6-4877-a7ec-11ac5ecdee6f · outbound

This paper cites Attention Heads of Large Language Models: A Survey.

CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation Attention Heads of Large Language Models: A Survey

Reference 33

Resolution: unresolved; no resolver link, observed 2026-08-06T05:00:04.960649Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation 74ef57b3-bd3a-4e58-9d51-25808beb877d · inbound

TPLA: Tensor Parallel Latent Attention for Efficient Disaggregated Prefill and Decode Inference cites this paper.

TPLA: Tensor Parallel Latent Attention for Efficient Disaggregated Prefill and Decode Inference CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation

Reference 23

Resolution: unresolved; no resolver link, observed 2026-08-05T17:55:08.361255Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3a5817d4-406b-4e47-aac6-a9ec47d5c553 · inbound

RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention cites this paper.

RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation

Reference 22

Resolution: verified exact; arxiv_id, observed 2026-07-02T13:46:59.563761Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation b6c4184e-7440-4a1e-8d3f-f2fc5b4e473b · inbound

RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention cites this paper.

RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation

Reference 36

Resolution: verified exact; arxiv_id, observed 2026-06-29T18:23:51.286192Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation 73f9c723-cfc9-4014-964a-9d40153a4680 · inbound

RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention cites this paper.

RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation

Reference 35

Resolution: unresolved; no resolver link, observed 2026-08-02T12:21:12.951114Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7bd4a058-0963-4f14-8896-5d35a218da9a · inbound

Dustin: Draft-Augmented Sparse Verification for Efficient Long-Context Generation with Speculative Decoding cites this paper.

Dustin: Draft-Augmented Sparse Verification for Efficient Long-Context Generation with Speculative Decoding CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation

Reference 6

Resolution: verified exact; arxiv_id, observed 2026-07-04T16:49:58.012411Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation cb7f3924-44fd-4157-b2f2-e6892b24057f · inbound

Logit-Contribution Scoring Identifies Non-Literal Retrieval Heads cites this paper.

Logit-Contribution Scoring Identifies Non-Literal Retrieval Heads CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation

Reference 18

Resolution: verified exact; arxiv_id, observed 2026-07-02T12:56:56.481008Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-08T06:32:00.761636+00:00.

Observation c4c6b158-d9d8-4ea5-a3db-76de214a1547 · inbound

Adaptive Filtering of the KV Cache: Diagnosing and Correcting Structural-Role Bias in LLM Inference cites this paper.

Adaptive Filtering of the KV Cache: Diagnosing and Correcting Structural-Role Bias in LLM Inference CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-02T05:58:47.023358Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5cbfbc9c-57d0-48b7-a240-4400d564416c · inbound

Training-Free Hashing-Based Attention via Binary Principal Components cites this paper.

Training-Free Hashing-Based Attention via Binary Principal Components CompressKV: Semantic Retrieval Heads Know What Tokens are Not Important Before Generation

Reference 60

Resolution: unresolved; no resolver link, observed 2026-08-07T00:45:48.497849Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.