Invoke Interfaces Only When Needed: Adaptive Invocation for Large Language Models in Question Answering Article Swipe

PDF

YOU? · · 2025 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2505.02311

The collaborative paradigm of large and small language models (LMs) effectively balances performance and cost, yet its pivotal challenge lies in precisely pinpointing the moment of invocation when hallucinations arise in small LMs. Previous optimization efforts primarily focused on post-processing techniques, which were separate from the reasoning process of LMs, resulting in high computational costs and limited effectiveness. In this paper, we propose a practical invocation evaluation metric called AttenHScore, which calculates the accumulation and propagation of hallucinations during the generation process of small LMs, continuously amplifying potential reasoning errors. By dynamically adjusting the detection threshold, we achieve more accurate real-time invocation of large LMs. Additionally, considering the limited reasoning capacity of small LMs, we leverage uncertainty-aware knowledge reorganization to assist them better capture critical information from different text chunks. Extensive experiments reveal that our AttenHScore outperforms most baselines in enhancing real-time hallucination detection capabilities across multiple QA datasets, especially when addressing complex queries. Moreover, our strategies eliminate the need for additional model training and display flexibility in adapting to various transformer-based LMs.

Related Topics

Only The Brave (2017 Film)

Only Murders In The Building

God Only Knows

My One And Only (Film)

Only Fools And Horses

You Only Live Twice (Film)

For Your Eyes Only (Film)

Only Yesterday (1991 Film)

Only Murders In The Building Season 4

Only Lovers Left Alive

Only God Forgives

It's Only Rock 'N Roll

Concepts

No concepts available.

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2505.02311
PDF: https://arxiv.org/pdf/2505.02311
OA Status: green
OpenAlex ID: https://openalex.org/W4415031204

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4415031204

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2505.02311

Digital Object Identifier
Title: Invoke Interfaces Only When Needed: Adaptive Invocation for Large Language Models in Question Answering

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2025

Year of publication
Publication date: 2025-05-05

Full publication date if available
Authors: J. Zhao, Chunlai Zhou, Biao Qin

List of authors in order
Landing page: https://arxiv.org/abs/2505.02311

Publisher landing page
PDF URL: https://arxiv.org/pdf/2505.02311

Direct link to full text PDF
Open access: Yes

Whether a free full text is available
OA status: green

Open access status per OpenAlex
OA URL: https://arxiv.org/pdf/2505.02311

Direct OA link when available
Cited by: 0

Total citation count in OpenAlex

Full payload

id	https://openalex.org/W4415031204
doi	https://doi.org/10.48550/arxiv.2505.02311
ids.doi	https://doi.org/10.48550/arxiv.2505.02311
ids.openalex	https://openalex.org/W4415031204
fwci
type	preprint
title	Invoke Interfaces Only When Needed: Adaptive Invocation for Large Language Models in Question Answering
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T10028
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.9954000115394592
topics[0].domain.id	https://openalex.org/domains/3
topics[0].domain.display_name	Physical Sciences
topics[0].subfield.id	https://openalex.org/subfields/1702
topics[0].subfield.display_name	Artificial Intelligence
topics[0].display_name	Topic Modeling
topics[1].id	https://openalex.org/T12031
topics[1].field.id	https://openalex.org/fields/17
topics[1].field.display_name	Computer Science
topics[1].score	0.9532999992370605
topics[1].domain.id	https://openalex.org/domains/3
topics[1].domain.display_name	Physical Sciences
topics[1].subfield.id	https://openalex.org/subfields/1702
topics[1].subfield.display_name	Artificial Intelligence
topics[1].display_name	Speech and dialogue systems
topics[2].id	https://openalex.org/T10181
topics[2].field.id	https://openalex.org/fields/17
topics[2].field.display_name	Computer Science
topics[2].score	0.9337999820709229
topics[2].domain.id	https://openalex.org/domains/3
topics[2].domain.display_name	Physical Sciences
topics[2].subfield.id	https://openalex.org/subfields/1702
topics[2].subfield.display_name	Artificial Intelligence
topics[2].display_name	Natural Language Processing Techniques
is_xpac	False
apc_list
apc_paid
language	en
locations[0].id	pmh:oai:arXiv.org:2505.02311
locations[0].is_oa	True
locations[0].source.id	https://openalex.org/S4306400194
locations[0].source.issn
locations[0].source.type	repository
locations[0].source.is_oa	True
locations[0].source.issn_l
locations[0].source.is_core	False
locations[0].source.is_in_doaj	False
locations[0].source.display_name	arXiv (Cornell University)
locations[0].source.host_organization	https://openalex.org/I205783295
locations[0].source.host_organization_name	Cornell University
locations[0].source.host_organization_lineage	https://openalex.org/I205783295
locations[0].license
locations[0].pdf_url	https://arxiv.org/pdf/2505.02311
locations[0].version	submittedVersion
locations[0].raw_type	text
locations[0].license_id
locations[0].is_accepted	False
locations[0].is_published	False
locations[0].raw_source_name
locations[0].landing_page_url	http://arxiv.org/abs/2505.02311
locations[1].id	doi:10.48550/arxiv.2505.02311
locations[1].is_oa	True
locations[1].source.id	https://openalex.org/S4306400194
locations[1].source.issn
locations[1].source.type	repository
locations[1].source.is_oa	True
locations[1].source.issn_l
locations[1].source.is_core	False
locations[1].source.is_in_doaj	False
locations[1].source.display_name	arXiv (Cornell University)
locations[1].source.host_organization	https://openalex.org/I205783295
locations[1].source.host_organization_name	Cornell University
locations[1].source.host_organization_lineage	https://openalex.org/I205783295
locations[1].license
locations[1].pdf_url
locations[1].version
locations[1].raw_type	article
locations[1].license_id
locations[1].is_accepted	False
locations[1].is_published
locations[1].raw_source_name
locations[1].landing_page_url	https://doi.org/10.48550/arxiv.2505.02311
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5070851446
authorships[0].author.orcid	https://orcid.org/0000-0001-6408-2314
authorships[0].author.display_name	J. Zhao
authorships[0].author_position	first
authorships[0].raw_author_name	Zhao, Jihao
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5027514426
authorships[1].author.orcid	https://orcid.org/0000-0002-5687-2196
authorships[1].author.display_name	Chunlai Zhou
authorships[1].author_position	middle
authorships[1].raw_author_name	Zhou, Chunlai
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5111326441
authorships[2].author.orcid
authorships[2].author.display_name	Biao Qin
authorships[2].author_position	last
authorships[2].raw_author_name	Qin, Biao
authorships[2].is_corresponding	False
has_content.pdf	False
has_content.grobid_xml	False
is_paratext	False
open_access.is_oa	True
open_access.oa_url	https://arxiv.org/pdf/2505.02311
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-10T00:00:00
display_name	Invoke Interfaces Only When Needed: Adaptive Invocation for Large Language Models in Question Answering
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T10028
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.9954000115394592
primary_topic.domain.id	https://openalex.org/domains/3
primary_topic.domain.display_name	Physical Sciences
primary_topic.subfield.id	https://openalex.org/subfields/1702
primary_topic.subfield.display_name	Artificial Intelligence
primary_topic.display_name	Topic Modeling
cited_by_count	0
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2505.02311
best_oa_location.is_oa	True
best_oa_location.source.id	https://openalex.org/S4306400194
best_oa_location.source.issn
best_oa_location.source.type	repository
best_oa_location.source.is_oa	True
best_oa_location.source.issn_l
best_oa_location.source.is_core	False
best_oa_location.source.is_in_doaj	False
best_oa_location.source.display_name	arXiv (Cornell University)
best_oa_location.source.host_organization	https://openalex.org/I205783295
best_oa_location.source.host_organization_name	Cornell University
best_oa_location.source.host_organization_lineage	https://openalex.org/I205783295
best_oa_location.license
best_oa_location.pdf_url	https://arxiv.org/pdf/2505.02311
best_oa_location.version	submittedVersion
best_oa_location.raw_type	text
best_oa_location.license_id
best_oa_location.is_accepted	False
best_oa_location.is_published	False
best_oa_location.raw_source_name
best_oa_location.landing_page_url	http://arxiv.org/abs/2505.02311
primary_location.id	pmh:oai:arXiv.org:2505.02311
primary_location.is_oa	True
primary_location.source.id	https://openalex.org/S4306400194
primary_location.source.issn
primary_location.source.type	repository
primary_location.source.is_oa	True
primary_location.source.issn_l
primary_location.source.is_core	False
primary_location.source.is_in_doaj	False
primary_location.source.display_name	arXiv (Cornell University)
primary_location.source.host_organization	https://openalex.org/I205783295
primary_location.source.host_organization_name	Cornell University
primary_location.source.host_organization_lineage	https://openalex.org/I205783295
primary_location.license
primary_location.pdf_url	https://arxiv.org/pdf/2505.02311
primary_location.version	submittedVersion
primary_location.raw_type	text
primary_location.license_id
primary_location.is_accepted	False
primary_location.is_published	False
primary_location.raw_source_name
primary_location.landing_page_url	http://arxiv.org/abs/2505.02311
publication_date	2025-05-05
publication_year	2025
referenced_works_count	0
abstract_inverted_index.a	63
abstract_inverted_index.By	90
abstract_inverted_index.In	58
abstract_inverted_index.QA	147
abstract_inverted_index.in	20, 30, 51, 139, 167
abstract_inverted_index.of	3, 25, 48, 76, 82, 102, 111
abstract_inverted_index.on	38
abstract_inverted_index.to	119, 169
abstract_inverted_index.we	61, 96, 114
abstract_inverted_index.The	0
abstract_inverted_index.and	5, 13, 55, 74, 164
abstract_inverted_index.for	160
abstract_inverted_index.its	16
abstract_inverted_index.our	134, 155
abstract_inverted_index.the	23, 45, 72, 79, 93, 107, 158
abstract_inverted_index.yet	15
abstract_inverted_index.LMs,	49, 84, 113
abstract_inverted_index.LMs.	32, 104, 172
abstract_inverted_index.from	44, 126
abstract_inverted_index.high	52
abstract_inverted_index.lies	19
abstract_inverted_index.more	98
abstract_inverted_index.most	137
abstract_inverted_index.need	159
abstract_inverted_index.text	128
abstract_inverted_index.that	133
abstract_inverted_index.them	121
abstract_inverted_index.this	59
abstract_inverted_index.were	42
abstract_inverted_index.when	27, 150
abstract_inverted_index.(LMs)	9
abstract_inverted_index.arise	29
abstract_inverted_index.cost,	14
abstract_inverted_index.costs	54
abstract_inverted_index.large	4, 103
abstract_inverted_index.model	162
abstract_inverted_index.small	6, 31, 83, 112
abstract_inverted_index.which	41, 70
abstract_inverted_index.across	145
abstract_inverted_index.assist	120
abstract_inverted_index.better	122
abstract_inverted_index.called	68
abstract_inverted_index.during	78
abstract_inverted_index.metric	67
abstract_inverted_index.models	8
abstract_inverted_index.moment	24
abstract_inverted_index.paper,	60
abstract_inverted_index.reveal	132
abstract_inverted_index.achieve	97
abstract_inverted_index.capture	123
abstract_inverted_index.chunks.	129
abstract_inverted_index.complex	152
abstract_inverted_index.display	165
abstract_inverted_index.efforts	35
abstract_inverted_index.errors.	89
abstract_inverted_index.focused	37
abstract_inverted_index.limited	56, 108
abstract_inverted_index.pivotal	17
abstract_inverted_index.process	47, 81
abstract_inverted_index.propose	62
abstract_inverted_index.various	170
abstract_inverted_index.Previous	33
abstract_inverted_index.accurate	99
abstract_inverted_index.adapting	168
abstract_inverted_index.balances	11
abstract_inverted_index.capacity	110
abstract_inverted_index.critical	124
abstract_inverted_index.language	7
abstract_inverted_index.leverage	115
abstract_inverted_index.multiple	146
abstract_inverted_index.paradigm	2
abstract_inverted_index.queries.	153
abstract_inverted_index.separate	43
abstract_inverted_index.training	163
abstract_inverted_index.Extensive	130
abstract_inverted_index.Moreover,	154
abstract_inverted_index.adjusting	92
abstract_inverted_index.baselines	138
abstract_inverted_index.challenge	18
abstract_inverted_index.datasets,	148
abstract_inverted_index.detection	94, 143
abstract_inverted_index.different	127
abstract_inverted_index.eliminate	157
abstract_inverted_index.enhancing	140
abstract_inverted_index.knowledge	117
abstract_inverted_index.potential	87
abstract_inverted_index.practical	64
abstract_inverted_index.precisely	21
abstract_inverted_index.primarily	36
abstract_inverted_index.real-time	100, 141
abstract_inverted_index.reasoning	46, 88, 109
abstract_inverted_index.resulting	50
abstract_inverted_index.additional	161
abstract_inverted_index.addressing	151
abstract_inverted_index.amplifying	86
abstract_inverted_index.calculates	71
abstract_inverted_index.especially	149
abstract_inverted_index.evaluation	66
abstract_inverted_index.generation	80
abstract_inverted_index.invocation	26, 65, 101
abstract_inverted_index.strategies	156
abstract_inverted_index.threshold,	95
abstract_inverted_index.AttenHScore	135
abstract_inverted_index.considering	106
abstract_inverted_index.dynamically	91
abstract_inverted_index.effectively	10
abstract_inverted_index.experiments	131
abstract_inverted_index.flexibility	166
abstract_inverted_index.information	125
abstract_inverted_index.outperforms	136
abstract_inverted_index.performance	12
abstract_inverted_index.pinpointing	22
abstract_inverted_index.propagation	75
abstract_inverted_index.techniques,	40
abstract_inverted_index.AttenHScore,	69
abstract_inverted_index.accumulation	73
abstract_inverted_index.capabilities	144
abstract_inverted_index.continuously	85
abstract_inverted_index.optimization	34
abstract_inverted_index.Additionally,	105
abstract_inverted_index.collaborative	1
abstract_inverted_index.computational	53
abstract_inverted_index.hallucination	142
abstract_inverted_index.effectiveness.	57
abstract_inverted_index.hallucinations	28, 77
abstract_inverted_index.reorganization	118
abstract_inverted_index.post-processing	39
abstract_inverted_index.transformer-based	171
abstract_inverted_index.uncertainty-aware	116
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	3
citation_normalized_percentile