{"paper_id":"2512.05156","title":"Semantic Faithfulness and Entropy Production Measures to Tame Your LLM Demons and Manage Hallucinations","authors":["Igor Halperin"],"paper_date":"2025-12-04","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2512.05156v2","page_url":"https://thequant.space/flowcharts/semantic-faithfulness-and-entropy-production-measures-to-tam/","scores":{"empirical_rigor":6,"math_complexity":7.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes two new unsupervised metrics for evaluating LLM faithfulness: the Semantic Faithfulness (SF) metric based on KL divergence between topic transition matrices, and the Semantic Entropy Production (SEP) metric derived from stochastic thermodynamics.","quote":"We propose two new unsupervised metrics for faithfulness evaluation using insights from information theory and thermodynamics. Our semantic faithfulness (SF) metric quantifies faithfulness for any given QCA triplet by the Kullback-Leibler (KL) divergence between these matrices.","summary":"Evaluating faithfulness of Large Language Models (LLMs) to a given task is a complex challenge. We propose two new unsupervised metrics for faithfulness evaluation using insights from information theory and thermodynamics. Our approach treats an LLM as a bipartite information engine where hidden…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The LLM is treated as a black box with no access to internal activations, log‑probs, or other internal information.","type":"model","stated":"explicit","quote":"The black-box setting of our formulation implies that we do not have access to additional information such as internal activations, log-probs etc.","support":"full"},{"assumption":"All questions Qk are constructed as semantically equivalent paraphrases of an initial question Q.","type":"data","stated":"explicit","quote":"we assume a more general setting where all questions Qk are constructed as semantically equivalent paraphrases of a given initial question Q. As discussed in [9], prompt paraphrasing is useful to enrich the data and extract more meaningful topic representations.","support":"full"},{"assumption":"The only available data for analysis are the Question‑Context‑Answer triplets (Qk, C, Ak).","type":"data","stated":"explicit","quote":"In what follows, we assume that triplets (Qk, C, Ak) constitute the only available data for our analysis.","support":"full"},{"assumption":"Topic distributions over shared topics can be represented as probability distributions and modeled with transition matrices Q and A.","type":"model","stated":"explicit","quote":"We model Question-Context-Answer (QCA) triplets as probability distributions over shared topics. Topic transformations from $C$ to $Q$ and $A$ are modeled as transition matrices ${\\bf Q}$ and ${\\bf A}$ encoding the query goal and actual result, respectively.","support":"full"},{"assumption":"The convex optimization problem for jointly inferring Q and A has a unique global optimum and converges reliably.","type":"statistical","stated":"explicit","quote":"Our proposed faithfulness metric, which we call the Semantic Faithfulness (SF) score and denote as FS, is defined in terms of the minimal value of this KL divergence (4), obtained by jointly optimizing over A and Q subject to all constraints: FS := 1 1 + Dmin . Importantly, both optimization problems are convex with guaranteed convergence to global optima, as established by the Csisz´ar-Tusn´ady alternating","support":"full"},{"assumption":"Embedding and clustering of text into a fixed number of semantic topics (N=23) adequately captures the semantic structure needed for the metrics.","type":"other","stated":"explicit","quote":"Clustering: The joint embedding space was clustered into N = 23 semantic topics using the Upper-Bounded Deterministic Information Bottleneck (UDIB) algorithm [10] in a single clustering pass.","support":"full"}]},"datasets":{"status":"available","items":[{"name":"NVIDIA 2024 Annual Report Risk Factors (Form 10‑K) section","provider":"NVIDIA Corporation","kind":"text-news","access":"public","period":"Fiscal year 2024","universe":null,"quote":"The context for all QCA triplets was extracted from NVIDIA Corpora-tion’s fiscal year 2024 Annual Report (Form 10-K), specifically the comprehensive Risk Factors section (Item 1A).","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"unclear"},"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":"Both groups of questions show similar mean Semantic Faithfulness (Group A FS = 0.509, Group B FS = 0.503), indicating question type alone does not strongly determine faithfulness.","kind":"statistical","quote":"Both groups exhibit similar mean Semantic Faith-fulness (Group A: FS = 0.509, Group B: FS = 0.503), indicating that question type alone does not strongly determine faithfulness in this dataset.","support":"full"},{"finding":"Individual triplet FS values range from 0.472 to 0.577, suggesting specific question‑context pairings matter more than broad categories.","kind":"statistical","quote":"However, individual triplet variation within groups (overall FS range: [0.472, 0.577]) suggests that specific question-context pairings matter more than broad question categories.","support":"full"},{"finding":"Positive correlation between question entropy H(Q) and SF (Pearson r = 0.695, p = 0.026).","kind":"statistical","quote":"We observe a positive correlation between H(Q) and FS (Pearson r = 0.695, p = 0.026), see Fig. 2.","support":"full"},{"finding":"Mean Semantic Entropy Production across all triplets is 0.287 bits (range 0.076–0.771 bits), with Group A showing higher SEP (0.359 bits) than Group B (0.215 bits).","kind":"statistical","quote":"Group A exhibits higher SEP (mean = 0.359 bits) than Group B (mean = 0.215 bits), consistent with Group A’s higher system entropy change.","support":"partial"},{"finding":"Correlation between FS and SEP is negative (r = ‑0.612, p = 0.060).","kind":"statistical","quote":"The correlation between FS and SEP is r = −0.612 (p = 0.060), see Fig. 3.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":18,"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":true,"code_url":"https://github.com/ighalp/semantic-faithfulness-sdm","note":null},"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":2,"extracted_at":"2026-11-14T19:45:29Z"},"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/"}