# Long-document and Multi-hop Question Answering on 2WikiMultiHopQA Source: [Long-document and Multi-hop Question Answering on 2WikiMultiHopQA]() Collection index: [llms.txt]() Data updated: 2026-06-11T11:37:01.925Z Evaluation of model performance on the 2WikiMultiHopQA dataset for long-document and multi-hop question answering, reporting F1 and ROUGE-L scores. 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 | Long-document and Multi-hop Question Answering | | Dataset (including any stated split) | 2WikiMultiHopQA | | Metric | F1 Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 10 of 10 | | 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 - [F1 Score](): Higher is better; unit: PERCENT. - [ROUGE-L Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | F1 Score | Result date | Report paper | | --- | --- | --- | --- | | D2Latom | 50.39 | Not recorded | 6a2a4157d33dfb23e78395df | | D2Lraw | 49.84 | Not recorded | 6a2a4157d33dfb23e78395df | | D2Latom | 47.43 | Not recorded | 6a2a4157d33dfb23e78395df | | D2Lraw | 46.06 | Not recorded | 6a2a4157d33dfb23e78395df | | Base+ctx | 45.31 | Not recorded | 6a2a4157d33dfb23e78395df | | Doc2Atom | 39.62 | Not recorded | 6a2a4157d33dfb23e78395df | | Base+ctx | 37.69 | Not recorded | 6a2a4157d33dfb23e78395df | | D2Lckpt | 36.08 | Not recorded | 6a2a4157d33dfb23e78395df | | Doc2Atom | 31.74 | Not recorded | 6a2a4157d33dfb23e78395df | | D2Lckpt | 11.25 | Not recorded | 6a2a4157d33dfb23e78395df | ### D2Latom F1 Score: 50.39. Result ID: 6a2a9cbefbc373592c28e5b7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Backbone":"Gemma-2-2B-It","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. ### D2Lraw F1 Score: 49.84. Result ID: 6a2a9cbefbc373592c28e5b6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Backbone":"Gemma-2-2B-It","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. ### D2Latom F1 Score: 47.43. Result ID: 6a2a9cbefbc373592c28e5bc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Backbone":"Qwen3-4B-Instruct","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. ### D2Lraw F1 Score: 46.06. Result ID: 6a2a9cbefbc373592c28e5bb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Backbone":"Qwen3-4B-Instruct","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. ### Base+ctx F1 Score: 45.31. Result ID: 6a2a9cbefbc373592c28e5b9. [Reporting paper]() Recorded evaluation configuration: {"LLM Backbone":"Qwen3-4B-Instruct","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. ### Doc2Atom F1 Score: 39.62. Result ID: 6a2a9cbefbc373592c28e5b8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Backbone":"Gemma-2-2B-It","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. ### Base+ctx F1 Score: 37.69. Result ID: 6a2a9cbefbc373592c28e5b4. [Reporting paper]() Recorded evaluation configuration: {"LLM Backbone":"Gemma-2-2B-It","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. ### D2Lckpt F1 Score: 36.08. Result ID: 6a2a9cbefbc373592c28e5b5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Backbone":"Gemma-2-2B-It","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. ### Doc2Atom F1 Score: 31.74. Result ID: 6a2a9cbefbc373592c28e5bd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Backbone":"Qwen3-4B-Instruct","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. ### D2Lckpt F1 Score: 11.25. Result ID: 6a2a9cbefbc373592c28e5ba. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"LLM Backbone":"Qwen3-4B-Instruct","Evaluation Protocol":"In-domain"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.