# Information Retrieval on SciFact BEIR (NDCG@10) Source: [Information Retrieval on SciFact BEIR (NDCG@10)]() Collection index: [llms.txt]() Data updated: 2026-06-29T07:15:46.412Z Evaluation of Information Retrieval performance on the SciFact BEIR dataset, specifically measured using the NDCG@10 metric. 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 | Information Retrieval | | Dataset (including any stated split) | SciFact BEIR | | Metric | NDCG@10 | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 42 of 42 | | 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 - [NDCG@10](): Higher is better; unit: PERCENT. - [Delta NDCG@10](): Higher is better; unit: SCALAR. - [95% CI Width (NDCG@10)](): Lower is better; unit: SCALAR. ## Recorded results | Method | NDCG@10 | Result date | Report paper | | --- | --- | --- | --- | | DIVE | 99.36 | Not recorded | 6a0e91a4f984739ed6738c2b | | Search | 98.59 | Not recorded | 6a0e91a4f984739ed6738c2b | | SMEC | 92.97 | Not recorded | 6a0e91a4f984739ed6738c2b | | inf-v1 | 85.4 | Not recorded | 68fa7bc41e209131cea8a4c0 | | inf-v1-1.5b | 82.8 | Not recorded | 68fa7bc41e209131cea8a4c0 | | Matryoshka | 80.05 | Not recorded | 6a0e91a4f984739ed6738c2b | | bge-en-icl | 79.1 | Not recorded | 68fa7bc41e209131cea8a4c0 | | Qwen3-8B | 78.5 | Not recorded | 68fa7bc41e209131cea8a4c0 | | LENS-d4K | 78.4 | Not recorded | 68fa7bc41e209131cea8a4c0 | | Qwen3-4B | 78.3 | Not recorded | 68fa7bc41e209131cea8a4c0 | | gte-base-v1.5 | 76.8 | Not recorded | 68fa7bc41e209131cea8a4c0 | | gte-base | 76.2 | Not recorded | 68fa7bc41e209131cea8a4c0 | | bge-lg-v1.5 | 74.6 | Not recorded | 68fa7bc41e209131cea8a4c0 | | gte-large | 74.3 | Not recorded | 68fa7bc41e209131cea8a4c0 | | MUGI | 73.5 | Not recorded | 6a28f37190966ade919c1930 | | STORM | 72.8 | Not recorded | 6a28f37190966ade919c1930 | | GenQR | 72.6 | Not recorded | 6a28f37190966ade919c1930 | | opensearch-gte | 72.5 | Not recorded | 68fa7bc41e209131cea8a4c0 | | GenQRE | 72.5 | Not recorded | 6a28f37190966ade919c1930 | | LameR | 72.5 | Not recorded | 6a28f37190966ade919c1930 | | e5-base-v2 + SOFTCENTROID | 72.41 | Not recorded | 6a0405b6dbc836b87d82768a | | bge-large | 72.4 | Not recorded | 68fa7bc41e209131cea8a4c0 | | CSQE | 72.1 | Not recorded | 6a28f37190966ade919c1930 | | e5-base-v2 | 72.04 | Not recorded | 6a0405b6dbc836b87d82768a | | Q2D | 72 | Not recorded | 6a28f37190966ade919c1930 | | e5-large-inst | 71.6 | Not recorded | 68fa7bc41e209131cea8a4c0 | | Splade-V3 | 71 | Not recorded | 68fa7bc41e209131cea8a4c0 | | Q2K | 70.9 | Not recorded | 6a28f37190966ade919c1930 | | QA-Expa | 70.6 | Not recorded | 6a28f37190966ade919c1930 | | e5-large | 70.2 | Not recorded | 68fa7bc41e209131cea8a4c0 | | MILCO | 70.1 | Not recorded | 68fa7bc41e209131cea8a4c0 | | Qwen3-0.6B | 69.7 | Not recorded | 68fa7bc41e209131cea8a4c0 | | LLM2Vec | 68.9 | Not recorded | 6a0e91a4f984739ed6738c2b | | RM3 | 64.6 | Not recorded | 6a28f37190966ade919c1930 | | raw | 63.7 | Not recorded | 6a41fbee206542251d7cca1f | | raw | 59.8 | Not recorded | 6a41fbee206542251d7cca1f | ### DIVE NDCG@10: 99.36. Result ID: 6a0eeb14fbc373592c071e75. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Dimension":"512d","Backbone":"LLM2Vec-Mistral-7B"}. This record does not assert independent reproduction. ### Search NDCG@10: 98.59. Result ID: 6a0eeb14fbc373592c071e73. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Dimension":"512d","Backbone":"LLM2Vec-Mistral-7B"}. This record does not assert independent reproduction. ### SMEC NDCG@10: 92.97. Result ID: 6a0eeb14fbc373592c071e74. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Dimension":"512d","Backbone":"LLM2Vec-Mistral-7B"}. This record does not assert independent reproduction. ### inf-v1 NDCG@10: 85.4. Result ID: 69bd171147f34ce924a7945e. [Reporting paper]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### inf-v1-1.5b NDCG@10: 82.8. Result ID: 69bd171147f34ce924a7945a. [Reporting paper]() Recorded evaluation configuration: {"Model Size Category":"\>=1B parameters"}. This record does not assert independent reproduction. ### Matryoshka NDCG@10: 80.05. Result ID: 6a0eeb14fbc373592c071e72. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Dimension":"512d","Backbone":"LLM2Vec-Mistral-7B"}. This record does not assert independent reproduction. ### bge-en-icl NDCG@10: 79.1. Result ID: 69bd171147f34ce924a79458. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size Category":"\>=1B parameters"}. This record does not assert independent reproduction. ### Qwen3-8B NDCG@10: 78.5. Result ID: 69bd171147f34ce924a79457. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size Category":"\>=1B parameters"}. This record does not assert independent reproduction. ### LENS-d4K NDCG@10: 78.4. Result ID: 69bd171147f34ce924a7945d. [Reporting paper]() Recorded evaluation configuration: {"Model Size Category":"\>=1B parameters"}. This record does not assert independent reproduction. ### Qwen3-4B NDCG@10: 78.3. Result ID: 69bd171147f34ce924a79459. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size Category":"\>=1B parameters"}. This record does not assert independent reproduction. ### gte-base-v1.5 NDCG@10: 76.8. Result ID: 69bd171147f34ce924a79461. [Reporting paper]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### gte-base NDCG@10: 76.2. Result ID: 69bd171147f34ce924a7945f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### bge-lg-v1.5 NDCG@10: 74.6. Result ID: 69bd171147f34ce924a79463. [Reporting paper]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### gte-large NDCG@10: 74.3. Result ID: 69bd171147f34ce924a7945c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size Category":"\>=1B parameters"}. This record does not assert independent reproduction. ### MUGI NDCG@10: 73.5. Result ID: 6a2a6ce5fbc373592c2763fb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4.1"}. This record does not assert independent reproduction. ### STORM NDCG@10: 72.8. Result ID: 6a2a6ce5fbc373592c2763fd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B","Max tokens":"64"}. This record does not assert independent reproduction. ### GenQR NDCG@10: 72.6. Result ID: 6a2a6ce5fbc373592c2763f5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4.1"}. This record does not assert independent reproduction. ### opensearch-gte NDCG@10: 72.5. Result ID: 69bd171147f34ce924a79465. [Reporting paper]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### GenQRE NDCG@10: 72.5. Result ID: 6a2a6ce5fbc373592c2763f6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4.1"}. This record does not assert independent reproduction. ### LameR NDCG@10: 72.5. Result ID: 6a2a6ce5fbc373592c2763fa. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"GPT-4.1"}. This record does not assert independent reproduction. ### e5-base-v2 + SOFTCENTROID NDCG@10: 72.41. Result ID: 6a04307547f34ce924e9ee95. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"e5-base-v2","Program Strategy":"SOFTCENTROID (default)","K":"3","α":"0.5","τ":"0.05","Cost":"1×","Per-task tuning":"No"}. This record does not assert independent reproduction. ### bge-large NDCG@10: 72.4. Result ID: 69bd171147f34ce924a79460. [Reporting paper]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### CSQE NDCG@10: 72.1. Result ID: 6a2a6ce5fbc373592c2763fc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4.1"}. This record does not assert independent reproduction. ### e5-base-v2 NDCG@10: 72.04. Result ID: 6a04307547f34ce924e9ee94. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model":"e5-base-v2"}. This record does not assert independent reproduction. ### Q2D NDCG@10: 72. Result ID: 6a2a6ce5fbc373592c2763f9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4.1","Evaluation Protocol":"Zero-shot"}. This record does not assert independent reproduction. ### e5-large-inst NDCG@10: 71.6. Result ID: 69bd171147f34ce924a7945b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size Category":"\>=1B parameters"}. This record does not assert independent reproduction. ### Splade-V3 NDCG@10: 71. Result ID: 69bd171147f34ce924a79466. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### Q2K NDCG@10: 70.9. Result ID: 6a2a6ce5fbc373592c2763f8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4.1"}. This record does not assert independent reproduction. ### QA-Expa NDCG@10: 70.6. Result ID: 6a2a6ce5fbc373592c2763f7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"GPT-4.1"}. This record does not assert independent reproduction. ### e5-large NDCG@10: 70.2. Result ID: 69bd171147f34ce924a79462. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### MILCO NDCG@10: 70.1. Result ID: 69bd171147f34ce924a79467. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### Qwen3-0.6B NDCG@10: 69.7. Result ID: 69bd171147f34ce924a79464. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Size Category":"\<1B parameters"}. This record does not assert independent reproduction. ### LLM2Vec NDCG@10: 68.9. Result ID: 6a0eeb14fbc373592c071e71. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Dimension":"4096d","Backbone":"LLM2Vec-Mistral-7B"}. This record does not assert independent reproduction. ### RM3 NDCG@10: 64.6. Result ID: 6a2a6ce5fbc373592c2763f4. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"GPT-4.1"}. This record does not assert independent reproduction. ### raw NDCG@10: 63.7. Result ID: 6a421b98fbc373592c8723af. [Reporting paper]() Recorded evaluation configuration: {"Encoder":"e5-base","Kcands":"200"}. This record does not assert independent reproduction. ### raw NDCG@10: 59.8. Result ID: 6a421b98fbc373592c8723b3. [Reporting paper]() Recorded evaluation configuration: {"Encoder":"e5-small","Kcands":"200"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.