# Multi-hop Question Answering on Complex-TR ODQA 1.0 (test) Source: [Multi-hop Question Answering on Complex-TR ODQA 1.0 (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:04:18.713Z Evaluation of model performance on the test split of the Complex-TR ODQA 1.0 dataset for Multi-hop Question Answering. 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 | Multi-hop Question Answering | | Dataset (including any stated split) | Complex-TR ODQA 1.0 (test) | | Metric | Set Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 13 of 13 | | 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 - [Set Accuracy](): Higher is better; unit: PERCENT. - [Answer F1 Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Set Accuracy | Result date | Report paper | | --- | --- | --- | --- | | FiD-PIT + Refine | 0.312 | Not recorded | 68fa740c1e209131cea5f719 | | FiD-PIT | 0.29 | Not recorded | 68fa740c1e209131cea5f719 | | FiD | 0.274 | Not recorded | 68fa740c1e209131cea5f719 | | PIT-SFT | 0.253 | Not recorded | 68fa740c1e209131cea5f719 | | FiD-PIT + Refine | 0.247 | Not recorded | 68fa740c1e209131cea5f719 | | FiD-PIT | 0.231 | Not recorded | 68fa740c1e209131cea5f719 | | SFT | 0.231 | Not recorded | 68fa740c1e209131cea5f719 | | PIT-SFT | 0.226 | Not recorded | 68fa740c1e209131cea5f719 | | SFT | 0.177 | Not recorded | 68fa740c1e209131cea5f719 | | FiD | 0.177 | Not recorded | 68fa740c1e209131cea5f719 | | FLAN-T5-XL | 0.145 | Not recorded | 68fa740c1e209131cea5f719 | | GPT-4 | 0.14 | Not recorded | 68fa740c1e209131cea5f719 | | GPT-3.5 | 0.097 | Not recorded | 68fa740c1e209131cea5f719 | ### FiD-PIT + Refine Set Accuracy: 0.312. Result ID: 699316253a6c314849139c91. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-large","Architecture":"FiD","Reasoning Enhancement":"PIT","Context Refinement":"true"}. This record does not assert independent reproduction. ### FiD-PIT Set Accuracy: 0.29. Result ID: 699316253a6c314849139c90. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-large","Architecture":"FiD","Reasoning Enhancement":"PIT"}. This record does not assert independent reproduction. ### FiD Set Accuracy: 0.274. Result ID: 699316253a6c314849139c8f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-large","Architecture":"FiD"}. This record does not assert independent reproduction. ### PIT-SFT Set Accuracy: 0.253. Result ID: 699316253a6c314849139c8e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-large","Architecture":"SFT","Reasoning Enhancement":"PIT"}. This record does not assert independent reproduction. ### FiD-PIT + Refine Set Accuracy: 0.247. Result ID: 699316253a6c314849139c8c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-base","Architecture":"FiD","Reasoning Enhancement":"PIT","Context Refinement":"true"}. This record does not assert independent reproduction. ### FiD-PIT Set Accuracy: 0.231. Result ID: 699316253a6c314849139c8b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-base","Architecture":"FiD","Reasoning Enhancement":"PIT"}. This record does not assert independent reproduction. ### SFT Set Accuracy: 0.231. Result ID: 699316253a6c314849139c8d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-large","Architecture":"SFT"}. This record does not assert independent reproduction. ### PIT-SFT Set Accuracy: 0.226. Result ID: 699316253a6c314849139c89. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-base","Architecture":"SFT","Reasoning Enhancement":"PIT"}. This record does not assert independent reproduction. ### SFT Set Accuracy: 0.177. Result ID: 699316253a6c314849139c88. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-base","Architecture":"SFT"}. This record does not assert independent reproduction. ### FiD Set Accuracy: 0.177. Result ID: 699316253a6c314849139c8a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"T5-base","Architecture":"FiD"}. This record does not assert independent reproduction. ### FLAN-T5-XL Set Accuracy: 0.145. Result ID: 699316253a6c314849139c85. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"FLAN-T5-XL"}. This record does not assert independent reproduction. ### GPT-4 Set Accuracy: 0.14. Result ID: 699316253a6c314849139c87. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"GPT-4"}. This record does not assert independent reproduction. ### GPT-3.5 Set Accuracy: 0.097. Result ID: 699316253a6c314849139c86. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Backbone":"GPT-3.5"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.