Paper Citation Record · LEDGER

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

As of 14 August 2026, this Paper Citation Record lists 20 of 20 outbound references and 12 inbound Pith citation observations for arXiv:2412.12094.

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

pith.paper-citation-record.v1
2412.12094 v6

Coverage vector

measured 20 of 20 reference resolution

Typed states for the displayed outbound observations.

Source: paper_references, paper_reference_links, observed 2026-08-11T14:25:06.182146Z

measured 32 of 32 standing notices

One-hop event checks from named stored sources.

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

measured 12 of 12 inbound itemization

Pith citing papers itemized under the disclosed page cap.

Source: paper_references, paper_reference_links, observed 2026-08-11T00:38:48.057219Z

measured 1 of 1 external citation measurements

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

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

Reference resolution

20 of 20 outbound references displayed

verified exact0
verified fuzzy7
unresolved13
parse uncertain0
malformed identifier0
metadata mismatch0

External citation measurements

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

Outbound references

Observation 0f0ff43e-151c-41d4-92bb-2a5d8e419bca · outbound

This paper cites The Falcon Series of Open Language Models.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator The Falcon Series of Open Language Models

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.023014Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2f865bea-b69d-4fb8-9556-181f5aa2d572 · outbound

This paper cites Additionally, feed-forward networks with ReLU activation can effectively represent any piecewise linear function.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator Additionally, feed-forward networks with ReLU activation can effectively represent any piecewise linear function

Reference 2

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T14:25:06.732832Z

Source-reported events for the cited work

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

Observation 5989b33f-4e24-40b9-a089-2a708f03521c · outbound

This paper cites The Pile: An 800GB Dataset of Diverse Text for Language Modeling.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator The Pile: An 800GB Dataset of Diverse Text for Language Modeling

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.063912Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 61972174-c251-47d3-a52a-ab1ebddcfd21 · outbound

This paper cites Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.085460Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a716a03f-1bf4-4da6-a402-865a87e69930 · outbound

This paper cites Chen, G., Xia, L., and Huang, C.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator Chen, G., Xia, L., and Huang, C

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.036755Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 18a6fb13-35f6-4da5-8002-6a5fb40ffaa9 · outbound

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

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funneling

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.100416Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 40a2fefc-d6f1-4bee-b9a2-6e587108eace · outbound

This paper cites an unresolved cited work.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator Unresolved cited work

Reference 13

Resolution: unresolved; raw_fallback, observed 2026-08-11T14:25:06.855705Z

Source-reported events for the cited work

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

Observation 788d97a0-b99e-46e8-9771-83c0ae06f2c4 · outbound

This paper cites an unresolved cited work.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator Unresolved cited work

Reference 14

Resolution: unresolved; raw_fallback, observed 2026-08-11T14:25:06.830308Z

Source-reported events for the cited work

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

Observation 79da381c-6552-44c3-b2ac-5d207f879b15 · outbound

This paper cites PyramidKV (Zhang et al., 2024).

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator PyramidKV (Zhang et al., 2024)

Reference 15

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T14:25:06.798914Z

Source-reported events for the cited work

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

Observation 0407bef2-3476-4576-a892-bfd3b702f81d · outbound

This paper cites ” and “?.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator ” and “?

Reference 16

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T14:25:06.776540Z

Source-reported events for the cited work

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

Observation eeb6ea4c-4082-4ae6-b4a7-3f7e7fa568a3 · outbound

This paper cites (n−1) d−1X i=0 δ−i :δ: (n−1) d−1X i=0 δ−i +δ −d+1 −δ # , u⊤Zk ∈.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator (n−1) d−1X i=0 δ−i :δ: (n−1) d−1X i=0 δ−i +δ −d+1 −δ # , u⊤Zk ∈

Reference 18

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T14:25:06.708598Z

Source-reported events for the cited work

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

Observation 4f79ad32-015f-4cf5-ad94-1a0d541c79ae · outbound

This paper cites 4 initial tokens are kept.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator 4 initial tokens are kept

Reference 19

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T14:25:06.674486Z

Source-reported events for the cited work

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

Observation 2ad808ad-71b5-426d-8d50-b350bd63c349 · outbound

This paper cites 32 initial tokens are kept.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator 32 initial tokens are kept

Reference 20

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T14:25:06.641891Z

Source-reported events for the cited work

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

Observation dcae64ab-e48c-4d7b-98f6-450c255989ef · outbound

This paper cites PyramidInfer: Pyramid KV Cache Compression for High-throughput LLM Inference.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator PyramidInfer: Pyramid KV Cache Compression for High-throughput LLM Inference

Reference 2017

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.072675Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a1039abf-a3ff-4e7b-ace6-453cbbbb4998 · outbound

This paper cites Training Verifiers to Solve Math Word Problems.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator Training Verifiers to Solve Math Word Problems

Reference 2018

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.049120Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 86b3e9f2-bd5e-4078-bcbe-a9110222a9d2 · outbound

This paper cites The Llama 3 Herd of Models.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator The Llama 3 Herd of Models

Reference 2020

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.057530Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fedc5b42-97d8-43f8-b498-74959d10411c · outbound

This paper cites The results in Table 17 show that FixLLM has a significant gap compared to SepLLM in both mathematical logical reasoning and knowledge-based reasoning capabili- ties.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator The results in Table 17 show that FixLLM has a significant gap compared to SepLLM in both mathematical logical reasoning and knowledge-based reasoning capabili- ties

Reference 2021

Resolution: verified fuzzy; raw_fallback, observed 2026-08-11T14:25:06.755414Z

Source-reported events for the cited work

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

Observation 9a80c46f-697b-45df-bcf6-9d1a3f52d890 · outbound

This paper cites Longformer: The Long-Document Transformer.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator Longformer: The Long-Document Transformer

Reference 2023

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.029478Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b0727550-c477-4fe7-a605-2f220933b70f · outbound

This paper cites Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge

Reference 2024

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.042723Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fe622c78-ea51-4154-a854-9c00766ff892 · outbound

This paper cites SampleAttention: Near-Lossless Acceleration of Long Context LLM Inference with Adaptive Structured Sparse Attention.

SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator SampleAttention: Near-Lossless Acceleration of Long Context LLM Inference with Adaptive Structured Sparse Attention

Reference 2025

Resolution: unresolved; no resolver link, observed 2026-08-11T14:25:06.112967Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Pith citing papers

Observation 71b2eebb-bf39-443f-b1bf-e96dcb0c4954 · inbound

A Survey on Large Language Model Acceleration based on KV Cache Management cites this paper.

A Survey on Large Language Model Acceleration based on KV Cache Management SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 136

Resolution: unresolved; no resolver link, observed 2026-08-11T00:38:48.057219Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation dc318184-6f14-4f85-aeb2-b1463aee789f · inbound

Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention cites this paper.

Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 51

Resolution: verified exact; arxiv_id, observed 2026-05-16T23:46:30.212281Z

Source-reported events for the cited work

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

Observation a5f70ddc-f502-4087-85e2-06912880a329 · inbound

GEM: Empowering LLM for both Embedding Generation and Language Understanding cites this paper.

GEM: Empowering LLM for both Embedding Generation and Language Understanding SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 26

Resolution: unresolved; no resolver link, observed 2026-08-07T10:50:51.014693Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8f496848-d855-4471-8142-2c892c297b6b · inbound

SeerAttention-R: Sparse Attention Adaptation for Long Reasoning cites this paper.

SeerAttention-R: Sparse Attention Adaptation for Long Reasoning SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-07T05:06:32.572234Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1a8464b1-0285-4290-bf96-5dec339b3a40 · inbound

EARN: Efficient Inference Acceleration for LLM-based Generative Recommendation by Register Tokens cites this paper.

EARN: Efficient Inference Acceleration for LLM-based Generative Recommendation by Register Tokens SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-06T21:18:46.625475Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7e3cbe33-ce91-4136-8db3-9c084ff0c855 · inbound

OrthoRank: Token Selection via Sink Token Orthogonality for Efficient LLM inference cites this paper.

OrthoRank: Token Selection via Sink Token Orthogonality for Efficient LLM inference SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-06T20:08:06.728594Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 57454bc2-d061-44aa-ae76-7f3e17e17fe3 · inbound

CaliDrop: KV Cache Compression with Calibration cites this paper.

CaliDrop: KV Cache Compression with Calibration SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-06T13:57:15.675892Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 97e2590e-7e60-4785-bac5-541b06db62e6 · inbound

ChunkLLM: A Lightweight Pluggable Framework for Accelerating LLMs Inference cites this paper.

ChunkLLM: A Lightweight Pluggable Framework for Accelerating LLMs Inference SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-04T14:43:00.999081Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4761a781-8aaa-4876-8e24-731d6cab01ea · inbound

LightThinker++: From Reasoning Compression to Memory Management cites this paper.

LightThinker++: From Reasoning Compression to Memory Management SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 23

Resolution: metadata mismatch; arxiv_id, observed 2026-05-13T17:28:01.786972Z

Source-reported events for the cited work

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

Observation 8ac373e4-d79f-4568-91fd-1f58389d87dd · inbound

SAGE: Selective Attention-Guided Extraction for Token-Efficient Document Indexing cites this paper.

SAGE: Selective Attention-Guided Extraction for Token-Efficient Document Indexing SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 2

Resolution: verified exact; arxiv_id, observed 2026-05-10T08:32:52.129872Z

Source-reported events for the cited work

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

Observation 8d5cab32-d60d-4afa-8d74-00ed049ff9fc · inbound

What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agents cites this paper.

What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agents SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 16

Resolution: verified exact; local_arxiv, observed 2026-07-10T01:36:44.291190Z

Source-reported events for the cited work

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

Observation ce8edf70-bd41-4f56-8b59-cb6c2af227a6 · inbound

Metaphor Tracer: A Theory-Informed Analysis of Hidden States cites this paper.

Metaphor Tracer: A Theory-Informed Analysis of Hidden States SepLLM: Accelerate Large Language Models by Compressing One Segment into One Separator

Reference 2024

Resolution: unresolved; no resolver link, observed 2026-07-31T07:30:05.476997Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.