# Hallucination Detection on HotpotQA (AUROC Step Analysis) Source: [Hallucination Detection on HotpotQA (AUROC Step Analysis)]() Collection index: [llms.txt]() Data updated: 2026-03-18T08:51:59.362Z Evaluation of model performance on detecting hallucinations within the HotpotQA dataset, specifically reporting AUROC results derived from 128 steps, 64 steps, and the average across steps. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Hallucination Detection | | Dataset (including any stated split) | HotpotQA | | Metric | AUROC (128 Steps) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 16 of 16 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [AUROC (128 Steps)](): Higher is better; unit: PERCENT. - [AUROC (64 Steps)](): Higher is better; unit: PERCENT. - [Average AUROC](): Higher is better; unit: PERCENT. ## Recorded results | Method | AUROC (128 Steps) | Result date | Report paper | | --- | --- | --- | --- | | DynHD | 84.2 | Not recorded | 69ba318b0636e5bf4afc0802 | | DynHD | 80.1 | Not recorded | 69ba318b0636e5bf4afc0802 | | TraceDet | 75.1 | Not recorded | 69ba318b0636e5bf4afc0802 | | TraceDet | 66.1 | Not recorded | 69ba318b0636e5bf4afc0802 | | TSV | 65 | Not recorded | 69ba318b0636e5bf4afc0802 | | EigenScore | 64.7 | Not recorded | 69ba318b0636e5bf4afc0802 | | Lexical Similarity | 64.2 | Not recorded | 69ba318b0636e5bf4afc0802 | | Semantic Entropy | 62.7 | Not recorded | 69ba318b0636e5bf4afc0802 | | EigenScore | 62.5 | Not recorded | 69ba318b0636e5bf4afc0802 | | Lexical Similarity | 59.7 | Not recorded | 69ba318b0636e5bf4afc0802 | | TSV | 58.7 | Not recorded | 69ba318b0636e5bf4afc0802 | | Semantic Entropy | 57.6 | Not recorded | 69ba318b0636e5bf4afc0802 | | CCS | 57.6 | Not recorded | 69ba318b0636e5bf4afc0802 | | LN-Entropy | 54.8 | Not recorded | 69ba318b0636e5bf4afc0802 | | CCS | 51.7 | Not recorded | 69ba318b0636e5bf4afc0802 | | Perplexity | 49.3 | Not recorded | 69ba318b0636e5bf4afc0802 | ### DynHD AUROC (128 Steps): 84.2. Result ID: 69ba66e347f34ce924a46622. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"LLaDA-8B-Instruct","Method Category":"Trajectory-based Methods"}. This record does not assert independent reproduction. ### DynHD AUROC (128 Steps): 80.1. Result ID: 69ba66e347f34ce924a46629. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Dream-7B-Instruct","Method Category":"Trajectory-based Methods"}. This record does not assert independent reproduction. ### TraceDet AUROC (128 Steps): 75.1. Result ID: 69ba66e347f34ce924a46628. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Dream-7B-Instruct","Method Category":"Trajectory-based Methods"}. This record does not assert independent reproduction. ### TraceDet AUROC (128 Steps): 66.1. Result ID: 69ba66e347f34ce924a46621. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"LLaDA-8B-Instruct","Method Category":"Trajectory-based Methods"}. This record does not assert independent reproduction. ### TSV AUROC (128 Steps): 65. Result ID: 69ba66e347f34ce924a46620. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"LLaDA-8B-Instruct","Method Category":"Latent-based Methods"}. This record does not assert independent reproduction. ### EigenScore AUROC (128 Steps): 64.7. Result ID: 69ba66e347f34ce924a4661e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"LLaDA-8B-Instruct","Method Category":"Latent-based Methods"}. This record does not assert independent reproduction. ### Lexical Similarity AUROC (128 Steps): 64.2. Result ID: 69ba66e347f34ce924a4661d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"LLaDA-8B-Instruct","Method Category":"Output-based Methods"}. This record does not assert independent reproduction. ### Semantic Entropy AUROC (128 Steps): 62.7. Result ID: 69ba66e347f34ce924a46623. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Dream-7B-Instruct","Method Category":"Output-based Methods"}. This record does not assert independent reproduction. ### EigenScore AUROC (128 Steps): 62.5. Result ID: 69ba66e347f34ce924a46625. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Dream-7B-Instruct","Method Category":"Latent-based Methods"}. This record does not assert independent reproduction. ### Lexical Similarity AUROC (128 Steps): 59.7. Result ID: 69ba66e347f34ce924a46624. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Dream-7B-Instruct","Method Category":"Output-based Methods"}. This record does not assert independent reproduction. ### TSV AUROC (128 Steps): 58.7. Result ID: 69ba66e347f34ce924a46627. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Dream-7B-Instruct","Method Category":"Latent-based Methods"}. This record does not assert independent reproduction. ### Semantic Entropy AUROC (128 Steps): 57.6. Result ID: 69ba66e347f34ce924a4661c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"LLaDA-8B-Instruct","Method Category":"Output-based Methods"}. This record does not assert independent reproduction. ### CCS AUROC (128 Steps): 57.6. Result ID: 69ba66e347f34ce924a4661f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"LLaDA-8B-Instruct","Method Category":"Latent-based Methods"}. This record does not assert independent reproduction. ### LN-Entropy AUROC (128 Steps): 54.8. Result ID: 69ba66e347f34ce924a4661b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"LLaDA-8B-Instruct","Method Category":"Output-based Methods"}. This record does not assert independent reproduction. ### CCS AUROC (128 Steps): 51.7. Result ID: 69ba66e347f34ce924a46626. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"Dream-7B-Instruct","Method Category":"Latent-based Methods"}. This record does not assert independent reproduction. ### Perplexity AUROC (128 Steps): 49.3. Result ID: 69ba66e347f34ce924a4661a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone Model":"LLaDA-8B-Instruct","Method Category":"Output-based Methods"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.