{"paper":{"title":"Fact4ac at the Financial Misinformation Detection Challenge Task: Reference-Free Financial Misinformation Detection via Fine-Tuning and Few-Shot Prompting of Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Fine-tuned LLMs detect financial misinformation at 95-96 percent accuracy using only internal context and no external references.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Cuong Hoang, Le-Minh Nguyen","submitted_at":"2026-04-16T05:38:58Z","abstract_excerpt":"The proliferation of financial misinformation poses a severe threat to market stability and investor trust, misleading market behavior and creating critical information asymmetry. Detecting such misleading narratives is inherently challenging, particularly in real-world scenarios where external evidence or supplementary references for cross-verification are strictly unavailable. This paper presents our winning methodology for the \"Reference-Free Financial Misinformation Detection\" shared task. Built upon the recently proposed RFC-BENCH framework (Jiang et al. 2026), this task challenges models"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our proposed system demonstrated superior efficacy, successfully securing the first-place ranking on both official leaderboards. Specifically, we achieved an accuracy of 95.4% on the public test set and 96.3% on the private test set.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the internal semantic understanding and contextual consistency of the fine-tuned LLMs are sufficient to determine the veracity of financial claims without any external evidence or references.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"LLM system with LoRA fine-tuning and few-shot prompting wins reference-free financial misinformation detection task at 95.4% public and 96.3% private accuracy.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Fine-tuned LLMs detect financial misinformation at 95-96 percent accuracy using only internal context and no external references.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"003213265fead6c24615fa81ad69e6de73c39d33e549ee6d58b316edeef63b87"},"source":{"id":"2604.14640","kind":"arxiv","version":1},"verdict":{"id":"d8bbd4a1-4def-4168-9045-402038739c84","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T12:04:31.096977Z","strongest_claim":"Our proposed system demonstrated superior efficacy, successfully securing the first-place ranking on both official leaderboards. Specifically, we achieved an accuracy of 95.4% on the public test set and 96.3% on the private test set.","one_line_summary":"LLM system with LoRA fine-tuning and few-shot prompting wins reference-free financial misinformation detection task at 95.4% public and 96.3% private accuracy.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the internal semantic understanding and contextual consistency of the fine-tuned LLMs are sufficient to determine the veracity of financial claims without any external evidence or references.","pith_extraction_headline":"Fine-tuned LLMs detect financial misinformation at 95-96 percent accuracy using only internal context and no external references."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.14640/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":5,"sample":[{"doi":"","year":2005,"title":"Language Models are Few-Shot Learners","work_id":"214732c0-2edd-44a0-af9e-28184a2b8279","ref_index":1,"cited_arxiv_id":"2005.14165","is_internal_anchor":true},{"doi":"","year":null,"title":"Meta-learning via language model in-context tuning.ArXiv, abs/2110.07814","work_id":"b1d50385-566e-45a8-aea2-67065ba7e5eb","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"InProceedings of the 33rd ACM Interna- tional Conference on Multimedia, MM ’25, 13874–13880","work_id":"983e2bf1-ece5-4ee0-9e95-482d4e1620e7","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"All that glisters is not gold: A bench- mark for reference-free counterfactual financial misinformation detection","work_id":"3e7cb2ba-4164-4df3-b06a-2cdcaba4bfeb","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Qwen2.5 Technical Report","work_id":"d8432992-4980-4a81-85c7-9fa2c2b87f85","ref_index":5,"cited_arxiv_id":"2412.15115","is_internal_anchor":true}],"resolved_work":5,"snapshot_sha256":"764a372c26735a6d47565ab4073bc24db6f57c012f47286bc496f7ae21132ab2","internal_anchors":2},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}