# Abstractive Summarization Faithfulness on XSum Source: [Abstractive Summarization Faithfulness on XSum]() Collection index: [llms.txt]() Data updated: 2026-06-04T07:28:39.810Z Evaluation of faithfulness metrics (R-L, BS-F1, FactKB) for abstractive text summarization models tested on the XSum dataset. 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 | Abstractive Summarization Faithfulness | | Dataset (including any stated split) | XSum | | Metric | R-L | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 17 of 17 | | 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 - [R-L](): Higher is better; unit: SCALAR. - [BS-F1](): Higher is better; unit: SCALAR. - [FactKB](): Higher is better; unit: SCALAR. ## Recorded results | Method | R-L | Result date | Report paper | | --- | --- | --- | --- | | CD | 22.71 | Not recorded | 6a210686c10484bac23f010c | | AD | 22.49 | Not recorded | 6a210686c10484bac23f010c | | DoLa (low) | 22.46 | Not recorded | 6a210686c10484bac23f010c | | DoLa (high) | 22.43 | Not recorded | 6a210686c10484bac23f010c | | Greedy | 22.41 | Not recorded | 6a210686c10484bac23f010c | | DCO | 22.4 | Not recorded | 6a210686c10484bac23f010c | | DeCoRe (static) | 21.94 | Not recorded | 6a210686c10484bac23f010c | | ITI | 21.64 | Not recorded | 6a210686c10484bac23f010c | | CAD | 21.45 | Not recorded | 6a210686c10484bac23f010c | | DCO | 20.39 | Not recorded | 6a210686c10484bac23f010c | | DoLa (high) | 19.92 | Not recorded | 6a210686c10484bac23f010c | | Greedy | 19.9 | Not recorded | 6a210686c10484bac23f010c | | DeCoRe (static) | 19.87 | Not recorded | 6a210686c10484bac23f010c | | DoLa (low) | 19.82 | Not recorded | 6a210686c10484bac23f010c | | AD | 19.79 | Not recorded | 6a210686c10484bac23f010c | | CAD | 18.82 | Not recorded | 6a210686c10484bac23f010c | | ITI | 13.25 | Not recorded | 6a210686c10484bac23f010c | ### CD R-L: 22.71. Result ID: 6a212925fbc373592c1c5130. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Llama-3-70B"}. This record does not assert independent reproduction. ### AD R-L: 22.49. Result ID: 6a212925fbc373592c1c5134. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Llama-3-70B"}. This record does not assert independent reproduction. ### DoLa (low) R-L: 22.46. Result ID: 6a212925fbc373592c1c5132. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-70B"}. This record does not assert independent reproduction. ### DoLa (high) R-L: 22.43. Result ID: 6a212925fbc373592c1c5133. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-70B"}. This record does not assert independent reproduction. ### Greedy R-L: 22.41. Result ID: 6a212925fbc373592c1c512e. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama-3-70B"}. This record does not assert independent reproduction. ### DCO R-L: 22.4. Result ID: 6a212925fbc373592c1c5136. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-70B"}. This record does not assert independent reproduction. ### DeCoRe (static) R-L: 21.94. Result ID: 6a212925fbc373592c1c5135. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-70B"}. This record does not assert independent reproduction. ### ITI R-L: 21.64. Result ID: 6a212925fbc373592c1c512f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-70B"}. This record does not assert independent reproduction. ### CAD R-L: 21.45. Result ID: 6a212925fbc373592c1c5131. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-70B"}. This record does not assert independent reproduction. ### DCO R-L: 20.39. Result ID: 6a212925fbc373592c1c512d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-8B"}. This record does not assert independent reproduction. ### DoLa (high) R-L: 19.92. Result ID: 6a212925fbc373592c1c512a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-8B"}. This record does not assert independent reproduction. ### Greedy R-L: 19.9. Result ID: 6a212925fbc373592c1c5126. [Reporting paper]() Recorded evaluation configuration: {"Model":"Llama-3-8B"}. This record does not assert independent reproduction. ### DeCoRe (static) R-L: 19.87. Result ID: 6a212925fbc373592c1c512c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-8B"}. This record does not assert independent reproduction. ### DoLa (low) R-L: 19.82. Result ID: 6a212925fbc373592c1c5129. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-8B"}. This record does not assert independent reproduction. ### AD R-L: 19.79. Result ID: 6a212925fbc373592c1c512b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"Llama-3-8B"}. This record does not assert independent reproduction. ### CAD R-L: 18.82. Result ID: 6a212925fbc373592c1c5128. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-8B"}. This record does not assert independent reproduction. ### ITI R-L: 13.25. Result ID: 6a212925fbc373592c1c5127. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Llama-3-8B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.