# H2 Assessment on HONEST Source: [H2 Assessment on HONEST]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:39:58.431Z Evaluation of model performance on the HONEST dataset using the H2 Assessment framework, focusing on scores related to poor, medium, and excellent performance ranges. 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 | H2 Assessment | | Dataset (including any stated split) | HONEST | | Metric | Count (1-3 Poor) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 18 of 18 | | 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 - [Count (1-3 Poor)](): Higher is better; unit: PERCENT. - [Count (4-6 Medium)](): Higher is better; unit: PERCENT. - [Count (7-10 Excellent)](): Higher is better; unit: PERCENT. - [H2 Score](): Higher is better; unit: SCALAR. - [Gain](): Higher is better; unit: SCALAR. ## Recorded results | Method | Count (1-3 Poor) | Result date | Report paper | | --- | --- | --- | --- | | Mistral-7b | 0.552 | Not recorded | 68fa762e1e209131cea6bb59 | | Llama2-7b | 0.429 | Not recorded | 68fa762e1e209131cea6bb59 | | Llama2-13b | 0.427 | Not recorded | 68fa762e1e209131cea6bb59 | | Llama2-70b | 0.394 | Not recorded | 68fa762e1e209131cea6bb59 | | ChatGPT | 0.385 | Not recorded | 68fa762e1e209131cea6bb59 | | Mixtral-8x7b | 0.314 | Not recorded | 68fa762e1e209131cea6bb59 | | Llama3-70b | 0.253 | Not recorded | 68fa762e1e209131cea6bb59 | | Llama2-13b | 0.249 | Not recorded | 68fa762e1e209131cea6bb59 | | Llama2-7b | 0.232 | Not recorded | 68fa762e1e209131cea6bb59 | | Mistral-7b | 0.217 | Not recorded | 68fa762e1e209131cea6bb59 | | Llama2-70b | 0.21 | Not recorded | 68fa762e1e209131cea6bb59 | | Claude3-Opus | 0.144 | Not recorded | 68fa762e1e209131cea6bb59 | | ChatGPT | 0.111 | Not recorded | 68fa762e1e209131cea6bb59 | | Llama3-70b | 0.042 | Not recorded | 68fa762e1e209131cea6bb59 | | Mixtral-8x7b | 0.028 | Not recorded | 68fa762e1e209131cea6bb59 | | GPT4 | 0.025 | Not recorded | 68fa762e1e209131cea6bb59 | | Claude3-Opus | 0.009 | Not recorded | 68fa762e1e209131cea6bb59 | | GPT4 | 0.001 | Not recorded | 68fa762e1e209131cea6bb59 | ### Mistral-7b Count (1-3 Poor): 0.552. Result ID: 69942a723a6c3148491b81c7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"raw"}. This record does not assert independent reproduction. ### Llama2-7b Count (1-3 Poor): 0.429. Result ID: 69942a723a6c3148491b81cb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"raw"}. This record does not assert independent reproduction. ### Llama2-13b Count (1-3 Poor): 0.427. Result ID: 69942a723a6c3148491b81cd. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"raw"}. This record does not assert independent reproduction. ### Llama2-70b Count (1-3 Poor): 0.394. Result ID: 69942a723a6c3148491b81cf. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"raw"}. This record does not assert independent reproduction. ### ChatGPT Count (1-3 Poor): 0.385. Result ID: 69942a723a6c3148491b81c3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"raw"}. This record does not assert independent reproduction. ### Mixtral-8x7b Count (1-3 Poor): 0.314. Result ID: 69942a723a6c3148491b81c9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"raw"}. This record does not assert independent reproduction. ### Llama3-70b Count (1-3 Poor): 0.253. Result ID: 69942a723a6c3148491b81d1. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"raw"}. This record does not assert independent reproduction. ### Llama2-13b Count (1-3 Poor): 0.249. Result ID: 69942a723a6c3148491b81ce. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"curiosity-driven prompting"}. This record does not assert independent reproduction. ### Llama2-7b Count (1-3 Poor): 0.232. Result ID: 69942a723a6c3148491b81cc. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"curiosity-driven prompting"}. This record does not assert independent reproduction. ### Mistral-7b Count (1-3 Poor): 0.217. Result ID: 69942a723a6c3148491b81c8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"curiosity-driven prompting"}. This record does not assert independent reproduction. ### Llama2-70b Count (1-3 Poor): 0.21. Result ID: 69942a723a6c3148491b81d0. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"curiosity-driven prompting"}. This record does not assert independent reproduction. ### Claude3-Opus Count (1-3 Poor): 0.144. Result ID: 69942a723a6c3148491b81c5. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"raw"}. This record does not assert independent reproduction. ### ChatGPT Count (1-3 Poor): 0.111. Result ID: 69942a723a6c3148491b81c4. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"curiosity-driven prompting"}. This record does not assert independent reproduction. ### Llama3-70b Count (1-3 Poor): 0.042. Result ID: 69942a723a6c3148491b81d2. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"curiosity-driven prompting"}. This record does not assert independent reproduction. ### Mixtral-8x7b Count (1-3 Poor): 0.028. Result ID: 69942a723a6c3148491b81ca. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"curiosity-driven prompting"}. This record does not assert independent reproduction. ### GPT4 Count (1-3 Poor): 0.025. Result ID: 69942a723a6c3148491b81c1. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"raw"}. This record does not assert independent reproduction. ### Claude3-Opus Count (1-3 Poor): 0.009. Result ID: 69942a723a6c3148491b81c6. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"curiosity-driven prompting"}. This record does not assert independent reproduction. ### GPT4 Count (1-3 Poor): 0.001. Result ID: 69942a723a6c3148491b81c2. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Optimization":"curiosity-driven prompting"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.