From RAGs to rich parameters: Probing how language models utilize external knowledge over parametric information for factual queries Article Swipe
Hitesh Wadhwa
,
Rahul Seetharaman
,
Somyaa Aggarwal
,
Reshmi Ghosh
,
Samyadeep Basu
,
Soundararajan Srinivasan
,
Wenlong Zhao
,
Shreyas Chaudhari
,
Ehsan Aghazadeh
·
YOU?
·
· 2024
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2406.12824
YOU?
·
· 2024
· Open Access
·
· DOI: https://doi.org/10.48550/arxiv.2406.12824
Retrieval Augmented Generation (RAG) enriches the ability of language models to reason using external context to augment responses for a given user prompt. This approach has risen in popularity due to practical applications in various applications of language models in search, question/answering, and chat-bots. However, the exact nature of how this approach works isn't clearly understood. In this paper, we mechanistically examine the RAG pipeline to highlight that language models take shortcut and have a strong bias towards utilizing only the context information to answer the question, while relying minimally on their parametric memory. We probe this mechanistic behavior in language models with: (i) Causal Mediation Analysis to show that the parametric memory is minimally utilized when answering a question and (ii) Attention Contributions and Knockouts to show that the last token residual stream do not get enriched from the subject token in the question, but gets enriched from other informative tokens in the context. We find this pronounced shortcut behaviour true across both LLaMa and Phi family of models.
Related Topics
Metadata
- Type
- preprint
- Language
- en
- Landing Page
- http://arxiv.org/abs/2406.12824
- https://arxiv.org/pdf/2406.12824
- OA Status
- green
- Cited By
- 2
- Related Works
- 10
- OpenAlex ID
- https://openalex.org/W4399838049
All OpenAlex metadata
Raw OpenAlex JSON
- OpenAlex ID
-
https://openalex.org/W4399838049Canonical identifier for this work in OpenAlex
- DOI
-
https://doi.org/10.48550/arxiv.2406.12824Digital Object Identifier
- Title
-
From RAGs to rich parameters: Probing how language models utilize external knowledge over parametric information for factual queriesWork title
- Type
-
preprintOpenAlex work type
- Language
-
enPrimary language
- Publication year
-
2024Year of publication
- Publication date
-
2024-06-18Full publication date if available
- Authors
-
Hitesh Wadhwa, Rahul Seetharaman, Somyaa Aggarwal, Reshmi Ghosh, Samyadeep Basu, Soundararajan Srinivasan, Wenlong Zhao, Shreyas Chaudhari, Ehsan AghazadehList of authors in order
- Landing page
-
https://arxiv.org/abs/2406.12824Publisher landing page
- PDF URL
-
https://arxiv.org/pdf/2406.12824Direct link to full text PDF
- Open access
-
YesWhether a free full text is available
- OA status
-
greenOpen access status per OpenAlex
- OA URL
-
https://arxiv.org/pdf/2406.12824Direct OA link when available
- Concepts
-
Parametric statistics, Computer science, Parametric model, Mathematics, StatisticsTop concepts (fields/topics) attached by OpenAlex
- Cited by
-
2Total citation count in OpenAlex
- Citations by year (recent)
-
2025: 1, 2024: 1Per-year citation counts (last 5 years)
- Related works (count)
-
10Other works algorithmically related by OpenAlex
Full payload
| id | https://openalex.org/W4399838049 |
|---|---|
| doi | https://doi.org/10.48550/arxiv.2406.12824 |
| ids.doi | https://doi.org/10.48550/arxiv.2406.12824 |
| ids.openalex | https://openalex.org/W4399838049 |
| fwci | 0.0 |
| type | preprint |
| title | From RAGs to rich parameters: Probing how language models utilize external knowledge over parametric information for factual queries |
| biblio.issue | |
| biblio.volume | |
| biblio.last_page | |
| biblio.first_page | |
| topics[0].id | https://openalex.org/T10181 |
| topics[0].field.id | https://openalex.org/fields/17 |
| topics[0].field.display_name | Computer Science |
| topics[0].score | 0.9825000166893005 |
| 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 | Natural Language Processing Techniques |
| topics[1].id | https://openalex.org/T10028 |
| topics[1].field.id | https://openalex.org/fields/17 |
| topics[1].field.display_name | Computer Science |
| topics[1].score | 0.972000002861023 |
| 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 | Topic Modeling |
| topics[2].id | https://openalex.org/T10215 |
| topics[2].field.id | https://openalex.org/fields/17 |
| topics[2].field.display_name | Computer Science |
| topics[2].score | 0.9329000115394592 |
| 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 | Semantic Web and Ontologies |
| is_xpac | False |
| apc_list | |
| apc_paid | |
| concepts[0].id | https://openalex.org/C117251300 |
| concepts[0].level | 2 |
| concepts[0].score | 0.6831014156341553 |
| concepts[0].wikidata | https://www.wikidata.org/wiki/Q1849855 |
| concepts[0].display_name | Parametric statistics |
| concepts[1].id | https://openalex.org/C41008148 |
| concepts[1].level | 0 |
| concepts[1].score | 0.6414912343025208 |
| concepts[1].wikidata | https://www.wikidata.org/wiki/Q21198 |
| concepts[1].display_name | Computer science |
| concepts[2].id | https://openalex.org/C24574437 |
| concepts[2].level | 3 |
| concepts[2].score | 0.49291324615478516 |
| concepts[2].wikidata | https://www.wikidata.org/wiki/Q7135228 |
| concepts[2].display_name | Parametric model |
| concepts[3].id | https://openalex.org/C33923547 |
| concepts[3].level | 0 |
| concepts[3].score | 0.16614916920661926 |
| concepts[3].wikidata | https://www.wikidata.org/wiki/Q395 |
| concepts[3].display_name | Mathematics |
| concepts[4].id | https://openalex.org/C105795698 |
| concepts[4].level | 1 |
| concepts[4].score | 0.15860885381698608 |
| concepts[4].wikidata | https://www.wikidata.org/wiki/Q12483 |
| concepts[4].display_name | Statistics |
| keywords[0].id | https://openalex.org/keywords/parametric-statistics |
| keywords[0].score | 0.6831014156341553 |
| keywords[0].display_name | Parametric statistics |
| keywords[1].id | https://openalex.org/keywords/computer-science |
| keywords[1].score | 0.6414912343025208 |
| keywords[1].display_name | Computer science |
| keywords[2].id | https://openalex.org/keywords/parametric-model |
| keywords[2].score | 0.49291324615478516 |
| keywords[2].display_name | Parametric model |
| keywords[3].id | https://openalex.org/keywords/mathematics |
| keywords[3].score | 0.16614916920661926 |
| keywords[3].display_name | Mathematics |
| keywords[4].id | https://openalex.org/keywords/statistics |
| keywords[4].score | 0.15860885381698608 |
| keywords[4].display_name | Statistics |
| language | en |
| locations[0].id | pmh:oai:arXiv.org:2406.12824 |
| 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/2406.12824 |
| 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/2406.12824 |
| locations[1].id | doi:10.48550/arxiv.2406.12824 |
| 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 | cc-by |
| locations[1].pdf_url | |
| locations[1].version | |
| locations[1].raw_type | article |
| locations[1].license_id | https://openalex.org/licenses/cc-by |
| 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.2406.12824 |
| indexed_in | arxiv, datacite |
| authorships[0].author.id | https://openalex.org/A5099282601 |
| authorships[0].author.orcid | |
| authorships[0].author.display_name | Hitesh Wadhwa |
| authorships[0].author_position | first |
| authorships[0].raw_author_name | Wadhwa, Hitesh |
| authorships[0].is_corresponding | False |
| authorships[1].author.id | https://openalex.org/A5104019728 |
| authorships[1].author.orcid | |
| authorships[1].author.display_name | Rahul Seetharaman |
| authorships[1].author_position | middle |
| authorships[1].raw_author_name | Seetharaman, Rahul |
| authorships[1].is_corresponding | False |
| authorships[2].author.id | https://openalex.org/A5111234400 |
| authorships[2].author.orcid | |
| authorships[2].author.display_name | Somyaa Aggarwal |
| authorships[2].author_position | middle |
| authorships[2].raw_author_name | Aggarwal, Somyaa |
| authorships[2].is_corresponding | False |
| authorships[3].author.id | https://openalex.org/A5019507987 |
| authorships[3].author.orcid | https://orcid.org/0000-0002-1814-2133 |
| authorships[3].author.display_name | Reshmi Ghosh |
| authorships[3].author_position | middle |
| authorships[3].raw_author_name | Ghosh, Reshmi |
| authorships[3].is_corresponding | False |
| authorships[4].author.id | https://openalex.org/A5085795724 |
| authorships[4].author.orcid | |
| authorships[4].author.display_name | Samyadeep Basu |
| authorships[4].author_position | middle |
| authorships[4].raw_author_name | Basu, Samyadeep |
| authorships[4].is_corresponding | False |
| authorships[5].author.id | https://openalex.org/A5102163048 |
| authorships[5].author.orcid | |
| authorships[5].author.display_name | Soundararajan Srinivasan |
| authorships[5].author_position | middle |
| authorships[5].raw_author_name | Srinivasan, Soundararajan |
| authorships[5].is_corresponding | False |
| authorships[6].author.id | https://openalex.org/A5102691719 |
| authorships[6].author.orcid | |
| authorships[6].author.display_name | Wenlong Zhao |
| authorships[6].author_position | middle |
| authorships[6].raw_author_name | Zhao, Wenlong |
| authorships[6].is_corresponding | False |
| authorships[7].author.id | https://openalex.org/A5113258513 |
| authorships[7].author.orcid | |
| authorships[7].author.display_name | Shreyas Chaudhari |
| authorships[7].author_position | middle |
| authorships[7].raw_author_name | Chaudhari, Shreyas |
| authorships[7].is_corresponding | False |
| authorships[8].author.id | https://openalex.org/A5099282602 |
| authorships[8].author.orcid | |
| authorships[8].author.display_name | Ehsan Aghazadeh |
| authorships[8].author_position | last |
| authorships[8].raw_author_name | Aghazadeh, Ehsan |
| authorships[8].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/2406.12824 |
| open_access.oa_status | green |
| open_access.any_repository_has_fulltext | False |
| created_date | 2025-10-10T00:00:00 |
| display_name | From RAGs to rich parameters: Probing how language models utilize external knowledge over parametric information for factual queries |
| has_fulltext | False |
| is_retracted | False |
| updated_date | 2025-11-06T06:51:31.235846 |
| primary_topic.id | https://openalex.org/T10181 |
| primary_topic.field.id | https://openalex.org/fields/17 |
| primary_topic.field.display_name | Computer Science |
| primary_topic.score | 0.9825000166893005 |
| 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 | Natural Language Processing Techniques |
| related_works | https://openalex.org/W2289718384, https://openalex.org/W1995675544, https://openalex.org/W2509524819, https://openalex.org/W2012121796, https://openalex.org/W2068427817, https://openalex.org/W2952090425, https://openalex.org/W2538333368, https://openalex.org/W3127866798, https://openalex.org/W4294845631, https://openalex.org/W1518153952 |
| cited_by_count | 2 |
| counts_by_year[0].year | 2025 |
| counts_by_year[0].cited_by_count | 1 |
| counts_by_year[1].year | 2024 |
| counts_by_year[1].cited_by_count | 1 |
| locations_count | 2 |
| best_oa_location.id | pmh:oai:arXiv.org:2406.12824 |
| 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/2406.12824 |
| 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/2406.12824 |
| primary_location.id | pmh:oai:arXiv.org:2406.12824 |
| 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/2406.12824 |
| 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/2406.12824 |
| publication_date | 2024-06-18 |
| publication_year | 2024 |
| referenced_works_count | 0 |
| abstract_inverted_index.a | 19, 74, 118 |
| abstract_inverted_index.In | 56 |
| abstract_inverted_index.We | 94, 155 |
| abstract_inverted_index.do | 134 |
| abstract_inverted_index.in | 27, 33, 39, 99, 142, 152 |
| abstract_inverted_index.is | 113 |
| abstract_inverted_index.of | 7, 36, 48, 168 |
| abstract_inverted_index.on | 90 |
| abstract_inverted_index.to | 10, 15, 30, 65, 83, 107, 126 |
| abstract_inverted_index.we | 59 |
| abstract_inverted_index.(i) | 103 |
| abstract_inverted_index.Phi | 166 |
| abstract_inverted_index.RAG | 63 |
| abstract_inverted_index.and | 42, 72, 120, 124, 165 |
| abstract_inverted_index.but | 145 |
| abstract_inverted_index.due | 29 |
| abstract_inverted_index.for | 18 |
| abstract_inverted_index.get | 136 |
| abstract_inverted_index.has | 25 |
| abstract_inverted_index.how | 49 |
| abstract_inverted_index.not | 135 |
| abstract_inverted_index.the | 5, 45, 62, 80, 85, 110, 129, 139, 143, 153 |
| abstract_inverted_index.(ii) | 121 |
| abstract_inverted_index.This | 23 |
| abstract_inverted_index.bias | 76 |
| abstract_inverted_index.both | 163 |
| abstract_inverted_index.find | 156 |
| abstract_inverted_index.from | 138, 148 |
| abstract_inverted_index.gets | 146 |
| abstract_inverted_index.have | 73 |
| abstract_inverted_index.last | 130 |
| abstract_inverted_index.only | 79 |
| abstract_inverted_index.show | 108, 127 |
| abstract_inverted_index.take | 70 |
| abstract_inverted_index.that | 67, 109, 128 |
| abstract_inverted_index.this | 50, 57, 96, 157 |
| abstract_inverted_index.true | 161 |
| abstract_inverted_index.user | 21 |
| abstract_inverted_index.when | 116 |
| abstract_inverted_index.(RAG) | 3 |
| abstract_inverted_index.LLaMa | 164 |
| abstract_inverted_index.exact | 46 |
| abstract_inverted_index.given | 20 |
| abstract_inverted_index.isn't | 53 |
| abstract_inverted_index.other | 149 |
| abstract_inverted_index.probe | 95 |
| abstract_inverted_index.risen | 26 |
| abstract_inverted_index.their | 91 |
| abstract_inverted_index.token | 131, 141 |
| abstract_inverted_index.using | 12 |
| abstract_inverted_index.while | 87 |
| abstract_inverted_index.with: | 102 |
| abstract_inverted_index.works | 52 |
| abstract_inverted_index.Causal | 104 |
| abstract_inverted_index.across | 162 |
| abstract_inverted_index.answer | 84 |
| abstract_inverted_index.family | 167 |
| abstract_inverted_index.memory | 112 |
| abstract_inverted_index.models | 9, 38, 69, 101 |
| abstract_inverted_index.nature | 47 |
| abstract_inverted_index.paper, | 58 |
| abstract_inverted_index.reason | 11 |
| abstract_inverted_index.stream | 133 |
| abstract_inverted_index.strong | 75 |
| abstract_inverted_index.tokens | 151 |
| abstract_inverted_index.ability | 6 |
| abstract_inverted_index.augment | 16 |
| abstract_inverted_index.clearly | 54 |
| abstract_inverted_index.context | 14, 81 |
| abstract_inverted_index.examine | 61 |
| abstract_inverted_index.memory. | 93 |
| abstract_inverted_index.models. | 169 |
| abstract_inverted_index.prompt. | 22 |
| abstract_inverted_index.relying | 88 |
| abstract_inverted_index.search, | 40 |
| abstract_inverted_index.subject | 140 |
| abstract_inverted_index.towards | 77 |
| abstract_inverted_index.various | 34 |
| abstract_inverted_index.Analysis | 106 |
| abstract_inverted_index.However, | 44 |
| abstract_inverted_index.approach | 24, 51 |
| abstract_inverted_index.behavior | 98 |
| abstract_inverted_index.context. | 154 |
| abstract_inverted_index.enriched | 137, 147 |
| abstract_inverted_index.enriches | 4 |
| abstract_inverted_index.external | 13 |
| abstract_inverted_index.language | 8, 37, 68, 100 |
| abstract_inverted_index.pipeline | 64 |
| abstract_inverted_index.question | 119 |
| abstract_inverted_index.residual | 132 |
| abstract_inverted_index.shortcut | 71, 159 |
| abstract_inverted_index.utilized | 115 |
| abstract_inverted_index.Attention | 122 |
| abstract_inverted_index.Augmented | 1 |
| abstract_inverted_index.Knockouts | 125 |
| abstract_inverted_index.Mediation | 105 |
| abstract_inverted_index.Retrieval | 0 |
| abstract_inverted_index.answering | 117 |
| abstract_inverted_index.behaviour | 160 |
| abstract_inverted_index.highlight | 66 |
| abstract_inverted_index.minimally | 89, 114 |
| abstract_inverted_index.practical | 31 |
| abstract_inverted_index.question, | 86, 144 |
| abstract_inverted_index.responses | 17 |
| abstract_inverted_index.utilizing | 78 |
| abstract_inverted_index.Generation | 2 |
| abstract_inverted_index.chat-bots. | 43 |
| abstract_inverted_index.parametric | 92, 111 |
| abstract_inverted_index.popularity | 28 |
| abstract_inverted_index.pronounced | 158 |
| abstract_inverted_index.information | 82 |
| abstract_inverted_index.informative | 150 |
| abstract_inverted_index.mechanistic | 97 |
| abstract_inverted_index.understood. | 55 |
| abstract_inverted_index.applications | 32, 35 |
| abstract_inverted_index.Contributions | 123 |
| abstract_inverted_index.mechanistically | 60 |
| abstract_inverted_index.question/answering, | 41 |
| cited_by_percentile_year | |
| countries_distinct_count | 0 |
| institutions_distinct_count | 9 |
| citation_normalized_percentile |