{"paper_id":"2412.12148","title":"How to Choose a Threshold for an Evaluation Metric for Large Language Models","authors":["Bhaskarjit Sarmah","Mingshu Li","Jingrao Lyu","Sebastian Frank","Nathalia Castellanos","Stefano Pasquali","Dhagash Mehta"],"paper_date":"2024-12-10","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2412.12148v1","page_url":"https://thequant.space/flowcharts/how-to-choose-a-threshold-for-an-evaluation-metric-for-large/","scores":{"empirical_rigor":7,"math_complexity":4.5,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper proposes a step‑by‑step recipe for selecting a robust threshold for a given LLM evaluation metric.","quote":"we propose a step-by-step recipe for picking a threshold for a given LLM evaluation metric.","summary":"To ensure and monitor large language models (LLMs) reliably, various evaluation metrics have been proposed in the literature. However, there is little research on prescribing a methodology to identify a robust threshold on these metrics even though there are many serious implications of an…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The Z‑score method assumes the evaluation metric scores follow a normal distribution or that the sample is large enough for the Central Limit Theorem to apply.","type":"statistical","stated":"explicit","quote":"The Z‑score approach is valid when the data follows a normal distribution or when the sample size is sufficiently large for the Central Limit Theorem to ensure approximate normality."},{"assumption":"Thresholds for bimodal (or multimodal) distributed scores can be identified by locating the midpoint between two modes.","type":"statistical","stated":"implied","quote":"Determining a threshold for bimodal (or multimodal) distributed scores involves identifying the midpoint that effectively separates the two modes."}]},"datasets":{"status":"available","items":[{"name":"HaluBench","provider":null,"kind":"benchmark","access":"public","period":null,"universe":null,"quote":"we focus on a publicly available dataset called HaluBench"},{"name":"UpTrain","provider":null,"kind":"benchmark","access":"public","period":null,"universe":null,"quote":"For UpTrain, we define gpt-4o-mini as our LLM model and set the evaluation metric to FACTUAL ACCURACY for faithfulness evaluation."},{"name":"DeepEval","provider":null,"kind":"benchmark","access":"public","period":null,"universe":null,"quote":"For the DeepEval framework, we specify our LLM model as gpt-4o-mini and set include_reason = True"},{"name":"RAGAS","provider":null,"kind":"benchmark","access":"public","period":null,"universe":null,"quote":"In RAGAS, we used the default LLM hyperparameters such as temperature and top-p."}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"none"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"unclear","findings":[{"finding":"The KDE method performs reasonably well at lower confidence levels (e.g., 80%) but fails to align with desired thresholds as confidence levels increase.","kind":"performance","quote":"however, as the confidence level increases, the thresholds derived from the KDE method often fail to align with the desired levels."},{"finding":"Empirical recall aligns with desired class separation at low confidence levels but becomes overly conservative at higher levels, sometimes defaulting to zero.","kind":"performance","quote":"at higher confidence levels, the thresholds derived using this method exhibit overly conservative behavior, particularly for the DeepEval dataset."},{"finding":"All conformal prediction methods achieve coverage rates that converge to the specified confidence levels, confirming the framework’s validity.","kind":"statistical","quote":"confidence levels, validating the conformal prediction framework."},{"finding":"GAM and polynomial logistic regression consistently achieve the highest AUC‑ROC and average precision scores across datasets.","kind":"performance","quote":"Overall, the best scores are consistently achieved either by GAM or polynomial logistic regression, highlighting their effectiveness in this context."}],"reported_performance":null,"out_of_sample":{"method":"cross-validation","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":27,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":1,"extracted_at":"2026-11-14T17:50:18Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}