{"id":"https://openalex.org/W7124172941","doi":"https://doi.org/10.3390/computers15010055","title":"SteadyEval: Robust LLM Exam Graders via Adversarial Training and Distillation","display_name":"SteadyEval: Robust LLM Exam Graders via Adversarial Training and Distillation","publication_year":2026,"publication_date":"2026-01-14","ids":{"openalex":"https://openalex.org/W7124172941","doi":"https://doi.org/10.3390/computers15010055"},"language":"en","primary_location":{"id":"doi:10.3390/computers15010055","is_oa":true,"landing_page_url":"https://doi.org/10.3390/computers15010055","pdf_url":"https://www.mdpi.com/2073-431X/15/1/55/pdf?version=1768386149","source":{"id":"https://openalex.org/S4210228075","display_name":"Computers","issn_l":"2073-431X","issn":["2073-431X"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310310987","host_organization_name":"Multidisciplinary Digital Publishing Institute","host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Computers","raw_type":"journal-article"},"type":"article","indexed_in":["crossref","doaj"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://www.mdpi.com/2073-431X/15/1/55/pdf?version=1768386149","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5049418389","display_name":"C\u0103t\u0103lin Anghel","orcid":"https://orcid.org/0000-0002-1849-3072"},"institutions":[{"id":"https://openalex.org/I3122748864","display_name":"\"Dunarea de Jos\" University of Galati","ror":"https://ror.org/052sta926","country_code":"RO","type":"education","lineage":["https://openalex.org/I3122748864"]}],"countries":["RO"],"is_corresponding":true,"raw_author_name":"Catalin Anghel","raw_affiliation_strings":["Department of Computer Science and Information Technology, \u201cDun\u0103rea de Jos\u201d University of Galati, \u0218tiin\u021bei St. 2, 800146 Galati, Romania"],"raw_orcid":"https://orcid.org/0000-0002-1849-3072","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Information Technology, \u201cDun\u0103rea de Jos\u201d University of Galati, \u0218tiin\u021bei St. 2, 800146 Galati, Romania","institution_ids":["https://openalex.org/I3122748864"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5121517948","display_name":"Marian Viorel Craciun","orcid":null},"institutions":[{"id":"https://openalex.org/I3122748864","display_name":"\"Dunarea de Jos\" University of Galati","ror":"https://ror.org/052sta926","country_code":"RO","type":"education","lineage":["https://openalex.org/I3122748864"]}],"countries":["RO"],"is_corresponding":false,"raw_author_name":"Marian Viorel Craciun","raw_affiliation_strings":["Department of Computer Science and Information Technology, \u201cDun\u0103rea de Jos\u201d University of Galati, \u0218tiin\u021bei St. 2, 800146 Galati, Romania"],"raw_orcid":"https://orcid.org/0009-0006-9970-2556","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Information Technology, \u201cDun\u0103rea de Jos\u201d University of Galati, \u0218tiin\u021bei St. 2, 800146 Galati, Romania","institution_ids":["https://openalex.org/I3122748864"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103039137","display_name":"Adina Cocu","orcid":"https://orcid.org/0000-0003-0935-4713"},"institutions":[{"id":"https://openalex.org/I3122748864","display_name":"\"Dunarea de Jos\" University of Galati","ror":"https://ror.org/052sta926","country_code":"RO","type":"education","lineage":["https://openalex.org/I3122748864"]}],"countries":["RO"],"is_corresponding":true,"raw_author_name":"Adina Cocu","raw_affiliation_strings":["Department of Computer Science and Information Technology, \u201cDun\u0103rea de Jos\u201d University of Galati, \u0218tiin\u021bei St. 2, 800146 Galati, Romania"],"raw_orcid":"https://orcid.org/0000-0003-0935-4713","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Information Technology, \u201cDun\u0103rea de Jos\u201d University of Galati, \u0218tiin\u021bei St. 2, 800146 Galati, Romania","institution_ids":["https://openalex.org/I3122748864"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119223040","display_name":"Andreea Alexandra Anghel","orcid":"https://orcid.org/0009-0008-2537-6713"},"institutions":[{"id":"https://openalex.org/I3122748864","display_name":"\"Dunarea de Jos\" University of Galati","ror":"https://ror.org/052sta926","country_code":"RO","type":"education","lineage":["https://openalex.org/I3122748864"]}],"countries":["RO"],"is_corresponding":false,"raw_author_name":"Andreea Alexandra Anghel","raw_affiliation_strings":["Faculty of Automation, Computer Science, Electrical and Electronic Engineering, \u201cDun\u0103rea de Jos\u201d University of Galati, 800008 Galati, Romania"],"raw_orcid":"https://orcid.org/0009-0008-2537-6713","affiliations":[{"raw_affiliation_string":"Faculty of Automation, Computer Science, Electrical and Electronic Engineering, \u201cDun\u0103rea de Jos\u201d University of Galati, 800008 Galati, Romania","institution_ids":["https://openalex.org/I3122748864"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5077496241","display_name":"Adrian Istrate","orcid":"https://orcid.org/0000-0002-1317-6676"},"institutions":[{"id":"https://openalex.org/I3122748864","display_name":"\"Dunarea de Jos\" University of Galati","ror":"https://ror.org/052sta926","country_code":"RO","type":"education","lineage":["https://openalex.org/I3122748864"]}],"countries":["RO"],"is_corresponding":false,"raw_author_name":"Adrian Istrate","raw_affiliation_strings":["Department of Computer Science and Information Technology, \u201cDun\u0103rea de Jos\u201d University of Galati, \u0218tiin\u021bei St. 2, 800146 Galati, Romania"],"raw_orcid":"https://orcid.org/0000-0002-1317-6676","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Information Technology, \u201cDun\u0103rea de Jos\u201d University of Galati, \u0218tiin\u021bei St. 2, 800146 Galati, Romania","institution_ids":["https://openalex.org/I3122748864"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5049418389","https://openalex.org/A5103039137"],"corresponding_institution_ids":["https://openalex.org/I3122748864"],"apc_list":{"value":1800,"currency":"CHF","value_usd":2172},"apc_paid":{"value":1800,"currency":"CHF","value_usd":2172},"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.05524667,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"15","issue":"1","first_page":"55","last_page":"55"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.34369999170303345,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.34369999170303345,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.16120000183582306,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.04899999871850014,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/pipeline","display_name":"Pipeline (software)","score":0.7706000208854675},{"id":"https://openalex.org/keywords/baseline","display_name":"Baseline (sea)","score":0.6341999769210815},{"id":"https://openalex.org/keywords/grading","display_name":"Grading (engineering)","score":0.6136000156402588},{"id":"https://openalex.org/keywords/adversarial-system","display_name":"Adversarial system","score":0.569599986076355},{"id":"https://openalex.org/keywords/pipeline-transport","display_name":"Pipeline transport","score":0.49300000071525574},{"id":"https://openalex.org/keywords/filter","display_name":"Filter (signal processing)","score":0.4108999967575073},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.3571999967098236},{"id":"https://openalex.org/keywords/stability","display_name":"Stability (learning theory)","score":0.3391999900341034}],"concepts":[{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.7706000208854675},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6862999796867371},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.6341999769210815},{"id":"https://openalex.org/C2777286243","wikidata":"https://www.wikidata.org/wiki/Q5591926","display_name":"Grading (engineering)","level":2,"score":0.6136000156402588},{"id":"https://openalex.org/C37736160","wikidata":"https://www.wikidata.org/wiki/Q1801315","display_name":"Adversarial system","level":2,"score":0.569599986076355},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5196999907493591},{"id":"https://openalex.org/C175309249","wikidata":"https://www.wikidata.org/wiki/Q725864","display_name":"Pipeline transport","level":2,"score":0.49300000071525574},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.45559999346733093},{"id":"https://openalex.org/C106131492","wikidata":"https://www.wikidata.org/wiki/Q3072260","display_name":"Filter (signal processing)","level":2,"score":0.4108999967575073},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3571999967098236},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.3391999900341034},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.33469998836517334},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.329800009727478},{"id":"https://openalex.org/C102366305","wikidata":"https://www.wikidata.org/wiki/Q1097688","display_name":"Nonparametric statistics","level":2,"score":0.31040000915527344},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.28999999165534973},{"id":"https://openalex.org/C94124525","wikidata":"https://www.wikidata.org/wiki/Q912550","display_name":"Categorization","level":2,"score":0.27799999713897705},{"id":"https://openalex.org/C111219384","wikidata":"https://www.wikidata.org/wiki/Q6954384","display_name":"NIST","level":2,"score":0.27469998598098755},{"id":"https://openalex.org/C148524875","wikidata":"https://www.wikidata.org/wiki/Q6975395","display_name":"F1 score","level":2,"score":0.27070000767707825},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.26249998807907104},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.25690001249313354}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.3390/computers15010055","is_oa":true,"landing_page_url":"https://doi.org/10.3390/computers15010055","pdf_url":"https://www.mdpi.com/2073-431X/15/1/55/pdf?version=1768386149","source":{"id":"https://openalex.org/S4210228075","display_name":"Computers","issn_l":"2073-431X","issn":["2073-431X"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310310987","host_organization_name":"Multidisciplinary Digital Publishing Institute","host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Computers","raw_type":"journal-article"},{"id":"pmh:oai:doaj.org/article:d689aeca72294b468ff68140ddeb0aa3","is_oa":true,"landing_page_url":"https://doaj.org/article/d689aeca72294b468ff68140ddeb0aa3","pdf_url":null,"source":{"id":"https://openalex.org/S4306401280","display_name":"DOAJ (DOAJ: Directory of Open Access Journals)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Computers, Vol 15, Iss 1, p 55 (2026)","raw_type":"article"}],"best_oa_location":{"id":"doi:10.3390/computers15010055","is_oa":true,"landing_page_url":"https://doi.org/10.3390/computers15010055","pdf_url":"https://www.mdpi.com/2073-431X/15/1/55/pdf?version=1768386149","source":{"id":"https://openalex.org/S4210228075","display_name":"Computers","issn_l":"2073-431X","issn":["2073-431X"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310310987","host_organization_name":"Multidisciplinary Digital Publishing Institute","host_organization_lineage":["https://openalex.org/P4310310987"],"host_organization_lineage_names":["Multidisciplinary Digital Publishing Institute"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Computers","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7124172941.pdf","grobid_xml":"https://content.openalex.org/works/W7124172941.grobid-xml"},"referenced_works_count":25,"referenced_works":["https://openalex.org/W3198690080","https://openalex.org/W4389344855","https://openalex.org/W4389408896","https://openalex.org/W4390549805","https://openalex.org/W4392976364","https://openalex.org/W4402640336","https://openalex.org/W4402671506","https://openalex.org/W4403503813","https://openalex.org/W4404515031","https://openalex.org/W4404781628","https://openalex.org/W4404783401","https://openalex.org/W4405239924","https://openalex.org/W4406457934","https://openalex.org/W4408292186","https://openalex.org/W4410550984","https://openalex.org/W4410862854","https://openalex.org/W4411429801","https://openalex.org/W4412995610","https://openalex.org/W4413025843","https://openalex.org/W4415210618","https://openalex.org/W4415445809","https://openalex.org/W4415456690","https://openalex.org/W4415903672","https://openalex.org/W4416093678","https://openalex.org/W4417478959"],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"are":[4,261],"increasingly":[5],"used":[6],"as":[7,264],"rubric-guided":[8,58,109],"graders":[9],"for":[10,197],"short-answer":[11,110],"exams,":[12],"but":[13,253],"their":[14],"decisions":[15],"can":[16],"be":[17],"unstable":[18],"across":[19,146,182],"prompts":[20],"and":[21,81,98,115,125,131,149,177,194,268,272],"vulnerable":[22],"to":[23,48,212,220,255],"answer-side":[24,50,129,158],"prompt":[25,51,147,183],"injection.":[26],"In":[27],"this":[28,190],"paper,":[29],"we":[30,118,132],"study":[31],"SteadyEval,":[32],"a":[33,73,82,87,99,174,217,228,265],"guardrailed":[34,83,186,243],"exam-grading":[35,67],"pipeline":[36,75,84,187,244],"in":[37,85,112,162,201,216,231],"which":[38,54,86],"an":[39,167],"adversarially":[40],"trained":[41],"LoRA":[42],"filter":[43,89],"(SteadyEval-7B-deep)":[44,90],"preprocesses":[45],"student":[46,78],"answers":[47,79,124],"remove":[49],"injection,":[52],"after":[53],"the":[55,61,96,104,134,155,163,202,221,242],"original":[56],"Mistral-7B-Instruct":[57],"grader":[59,101],"assigns":[60,103],"final":[62,105],"score.":[63,106],"We":[64],"build":[65],"two":[66,108],"pipelines":[68],"on":[69,139,173,224,250],"top":[70],"of":[71,122,128,136,169],"Mistral-7B-Instruct:":[72],"baseline":[74,165,223],"that":[76,233,241],"scores":[77,215],"directly,":[80],"LoRA-based":[88],"first":[91],"removes":[92,189],"injection":[93],"content":[94],"from":[95],"answer":[97],"downstream":[100],"then":[102],"Using":[107],"datasets":[111],"machine":[113],"learning":[114],"computer":[116],"networking,":[117],"generate":[119],"grouped":[120],"families":[121],"clean":[123,225],"four":[126],"classes":[127],"attacks,":[130],"evaluate":[133],"impact":[135],"these":[137],"attacks":[138,159],"score":[140,180],"shifts,":[141],"attack":[142],"success":[143],"rates,":[144],"stability":[145],"variants,":[148],"alignment":[150],"with":[151,210,227],"human":[152,213,246],"graders.":[153],"On":[154],"pooled":[156],"dataset,":[157],"inflate":[160],"grades":[161],"unguarded":[164,222],"by":[166],"average":[168],"about":[170],"+1.2":[171],"points":[172],"1\u201310":[175],"scale,":[176],"substantially":[178],"increase":[179],"dispersion":[181],"variants.":[184],"The":[185],"largely":[188],"systematic":[191],"grade":[192],"inflation":[193],"reduces":[195],"instability":[196],"many":[198],"items,":[199,252],"especially":[200],"machine-learning":[203,251],"exam,":[204],"while":[205],"keeping":[206],"mean":[207],"absolute":[208],"error":[209],"respect":[211],"reference":[214],"similar":[218],"range":[219],"answers,":[226],"conservative":[229],"shift":[230],"networking":[232,257],"motivates":[234],"per-course":[235,270],"calibration.":[236],"Chief-panel":[237],"comparisons":[238],"further":[239],"show":[240],"tracks":[245],"grading":[247],"more":[248],"closely":[249],"tends":[254],"under-score":[256],"answers.":[258],"These":[259],"findings":[260],"best":[262],"interpreted":[263],"proof-of-concept":[266],"guardrail":[267],"require":[269],"validation":[271],"calibration":[273],"before":[274],"operational":[275],"use.":[276]},"counts_by_year":[],"updated_date":"2026-08-25T07:29:55.448023","created_date":"2026-01-15T00:00:00"}