SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2408.03675 · ACL · 2024

NACL: A General and Effective KV Cache Eviction Framework for LLMs at Inference Time

Yu Sun, Zhenyu Zhang, Shuohuan Wang, Hua Wu, Yilong Chen, Junyuan Shang, Tingwen Liu, Guoxia Wang, Dianhai Yu, Shiyao Cui

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 2 failures of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2408.03675.json · badge: /api/ran/2408.03675.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
PaddlePaddle/Research canonical (classifier label)
classifier basis: census
3 ran · 0 no recorded run
FunctionStatusWhere it lives
get_logger Ran
ran; our contract draft was wrong, not the code
PaddlePaddle/Research/CV/SMILE/finetune.py
code served (permissive licence) · get_code("833ccd510fac73d0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Logger. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import logging
import os
import sys
import tempfile

# Create a temporary directory for the log file
temp_dir = tempfile.mkdtemp()
log_file_path = os.path.join(temp_dir, "test.log")

ARGS = {
    "filename": log_file_path,
    "logger_name": "test_logger"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Logger — a Logger output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Logger output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_dict Ran
ran; our contract draft was wrong, not the code
PaddlePaddle/Research/CV/PWCNet/infer.py
code served (permissive licence) · get_code("ac63e9a19256d49f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pickle
import tempfile
import os

# Create a temporary file with pickled data
test_dict = {
    'key1': 'value1',
    'key2': [1, 2, 3],
    'key3': {'nested': 'dict'}
}

# Create a temporary pickle file
temp_dir = tempfile.gettempdir()
temp_filename = os.path.join(temp_dir, 'test_pickle_load.pkl')

with open(temp_filename, 'wb') as f:
    pickle.dump(test_dict, f)

ARGS = {
    'filename_': temp_filename
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
position_encoding_init Ran
ran, honoured the contract we drafted
PaddlePaddle/Research/NLP/ACL2019-JEMT/model.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from autosimtrans/SimulTransBaseline, is pointer only (licence: NONE) · get_code("1492225d0a37a248")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from autosimtrans/SimulTransBaseline, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 0 later) and 0 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
get_logger 2 other papers
2 dated earlier
1702.00758 (2017-02)
HashNet: Deep Learning to Hash by Continuation
2111.09883 (2021-11)
Swin Transformer V2: Scaling Up Capacity and Resolution
load_dict 2 other papers
2 dated earlier
2210.13373 (2022-10)
Local Metric Learning for Off-Policy Evaluation in Contextual Bandits with Continuous Actions
2312.05063 (2023-12)
Individualizing Glioma Radiotherapy Planning by Optimization of Data and Physics-Informed Discrete Loss
position_encoding_init 1 other paper
1 dated earlier
1810.08398 (2018-10)
STACL: Simultaneous Translation with Implicit Anticipation and Controllable Latency using Prefix-to-Prefix Framework

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Large Language Models (LLMs) have ignited an innovative surge of AI applications, marking a new era of exciting possibilities equipped with extended context windows. However, hosting these models is cost-prohibitive mainly due to the extensive memory consumption of KV Cache involving long-context modeling. Despite several works proposing to evict unnecessary tokens from the KV Cache, most of them rely on the biased local statistics of accumulated attention scores and report performance using unconvincing metric like perplexity on inadequate short-text evaluation. In this paper, we propose NACL, a general framework for long-context KV cache eviction that achieves more optimal and efficient eviction in a single operation during the encoding phase. Due to NACL's efficiency, we combine more accurate attention score statistics in PROXY-TOKENS EVICTION with the diversified random eviction strategy of RANDOM EVICTION, aiming to alleviate the issue of attention bias and enhance the robustness in maintaining pivotal tokens for long-context modeling tasks. Notably, our method significantly improves the performance on short-and long-text tasks by 80% and 76% respectively, reducing KV Cache by up to 5× with over 95% performance maintenance.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2408.03675")
get_code_for_paper("2408.03675")
have("2408.03675")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2408.03675.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2408.03675.svg)](https://syntology.ai/paper/2408.03675)

Connect an agent — have() is free.