# Long-context Question Answering on 2WikiMQA Fixed Chunk 2048 Source: [Long-context Question Answering on 2WikiMQA Fixed Chunk 2048]() Collection index: [llms.txt]() Data updated: 2026-03-06T09:56:50.127Z Evaluation of model performance on the 2WikiMQA dataset specifically using fixed context chunks of size 2048 for long-context Question Answering tasks. 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-context Question Answering | | Dataset (including any stated split) | 2WikiMQA Fixed Chunk 2048 | | Metric | QA Score | | 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 - [QA Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | QA Score | Result date | Report paper | | --- | --- | --- | --- | | Baseline | 52.53 | Not recorded | 69aa5f972d603e6de3e41a04 | | Our + Reorder | 51.76 | Not recorded | 69aa5f972d603e6de3e41a04 | | Baseline | 51.61 | Not recorded | 69aa5f972d603e6de3e41a04 | | Our | 50.89 | Not recorded | 69aa5f972d603e6de3e41a04 | | Our | 50.64 | Not recorded | 69aa5f972d603e6de3e41a04 | | Our + Reorder | 47.73 | Not recorded | 69aa5f972d603e6de3e41a04 | | Our | 46.35 | Not recorded | 69aa5f972d603e6de3e41a04 | | Baseline | 45.88 | Not recorded | 69aa5f972d603e6de3e41a04 | | Our + Reorder | 44.55 | Not recorded | 69aa5f972d603e6de3e41a04 | | CacheBlend | 44.17 | Not recorded | 69aa5f972d603e6de3e41a04 | | EPIC (15%) | 44.01 | Not recorded | 69aa5f972d603e6de3e41a04 | | No Recompute | 43.7 | Not recorded | 69aa5f972d603e6de3e41a04 | | EPIC (15%) | 43.21 | Not recorded | 69aa5f972d603e6de3e41a04 | | CacheBlend | 42.26 | Not recorded | 69aa5f972d603e6de3e41a04 | | CacheBlend | 41.31 | Not recorded | 69aa5f972d603e6de3e41a04 | | EPIC (15%) | 40.87 | Not recorded | 69aa5f972d603e6de3e41a04 | | No Recompute | 39.69 | Not recorded | 69aa5f972d603e6de3e41a04 | | No Recompute | 39.48 | Not recorded | 69aa5f972d603e6de3e41a04 | ### Baseline QA Score: 52.53. Result ID: 69aaa4d76dbb7075b954d36d. [Reporting paper]() Recorded evaluation configuration: {"Model":"ChatGLM","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### Our + Reorder QA Score: 51.76. Result ID: 69aaa4d76dbb7075b954d370. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"ChatGLM","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### Baseline QA Score: 51.61. Result ID: 69aaa4d76dbb7075b954d361. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### Our QA Score: 50.89. Result ID: 69aaa4d76dbb7075b954d363. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### Our QA Score: 50.64. Result ID: 69aaa4d76dbb7075b954d36f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"ChatGLM","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### Our + Reorder QA Score: 47.73. Result ID: 69aaa4d76dbb7075b954d364. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### Our QA Score: 46.35. Result ID: 69aaa4d76dbb7075b954d369. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaMA","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### Baseline QA Score: 45.88. Result ID: 69aaa4d76dbb7075b954d367. [Reporting paper]() Recorded evaluation configuration: {"Model":"LLaMA","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### Our + Reorder QA Score: 44.55. Result ID: 69aaa4d76dbb7075b954d36a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaMA","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### CacheBlend QA Score: 44.17. Result ID: 69aaa4d76dbb7075b954d365. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### EPIC (15%) QA Score: 44.01. Result ID: 69aaa4d76dbb7075b954d372. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"ChatGLM","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### No Recompute QA Score: 43.7. Result ID: 69aaa4d76dbb7075b954d36e. [Reporting paper]() Recorded evaluation configuration: {"Model":"ChatGLM","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### EPIC (15%) QA Score: 43.21. Result ID: 69aaa4d76dbb7075b954d366. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### CacheBlend QA Score: 42.26. Result ID: 69aaa4d76dbb7075b954d371. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"ChatGLM","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### CacheBlend QA Score: 41.31. Result ID: 69aaa4d76dbb7075b954d36b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaMA","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### EPIC (15%) QA Score: 40.87. Result ID: 69aaa4d76dbb7075b954d36c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"LLaMA","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### No Recompute QA Score: 39.69. Result ID: 69aaa4d76dbb7075b954d368. [Reporting paper]() Recorded evaluation configuration: {"Model":"LLaMA","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. ### No Recompute QA Score: 39.48. Result ID: 69aaa4d76dbb7075b954d362. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen","Setting":"Fixed Chunk (2048)"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.