# Natural Language Understanding on SuperGLUE (CB, COPA, BoolQ, WiC subset) Source: [Natural Language Understanding on SuperGLUE (CB, COPA, BoolQ, WiC subset)]() Collection index: [llms.txt]() Data updated: 2026-03-12T09:24:14.752Z Evaluation of Natural Language Understanding capabilities on the SuperGLUE benchmark, specifically reporting results for the CB, COPA, BoolQ, and WiC tasks using accuracy metrics. 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 | Natural Language Understanding | | Dataset (including any stated split) | SuperGLUE | | Metric | CB Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 32 of 32 | | 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 - [CB Accuracy](): Higher is better; unit: PERCENT. - [COPA Accuracy](): Higher is better; unit: PERCENT. - [BoolQ Accuracy](): Higher is better; unit: PERCENT. - [WiC Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | CB Accuracy | Result date | Report paper | | --- | --- | --- | --- | | ARMADA | 94.5 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 93.4 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 93 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 92.8 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 92.8 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 92.8 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 87.4 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 87.4 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 85.7 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 85.7 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 85.6 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 83.8 | Not recorded | 69b2488ed030084c22e2f8d4 | | BERT-base | 83.7 | Not recorded | 69b2488ed030084c22e2f8d4 | | undistilled | 83.7 | Not recorded | 69b2488ed030084c22e2f8d4 | | DeBERTa-v2-xxlarge | 81.9 | Not recorded | 69b2488ed030084c22e2f8d4 | | undistilled | 81.9 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 80.2 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 78.6 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 78.5 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 78.5 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 78.5 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 78.4 | Not recorded | 69b2488ed030084c22e2f8d4 | | OPT-1.3B | 78.2 | Not recorded | 69b2488ed030084c22e2f8d4 | | undistilled | 78.2 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 76.8 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 76.7 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 76.7 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 76.6 | Not recorded | 69b2488ed030084c22e2f8d4 | | BERT-6L | 75 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 75 | Not recorded | 69b2488ed030084c22e2f8d4 | | undistilled | 75 | Not recorded | 69b2488ed030084c22e2f8d4 | | ARMADA | 73.1 | Not recorded | 69b2488ed030084c22e2f8d4 | ### ARMADA CB Accuracy: 94.5. Result ID: 69b285886dbb7075b95d6ef0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"DeBERTa-v2-xxlarge","Loss":"Lcosine","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 93.4. Result ID: 69b285886dbb7075b95d6ef1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"DeBERTa-v2-xxlarge","Loss":"Leuclid","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 93. Result ID: 69b285886dbb7075b95d6f12. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"DeBERTa-v2-xxlarge","Distillation type":"ARMADA - Lelementwise"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 92.8. Result ID: 69b285886dbb7075b95d6ef2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"DeBERTa-v2-xxlarge","Loss":"Lelementwise","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 92.8. Result ID: 69b285886dbb7075b95d6f10. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"DeBERTa-v2-xxlarge","Distillation type":"ARMADA - Lcosine"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 92.8. Result ID: 69b285886dbb7075b95d6f11. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"DeBERTa-v2-xxlarge","Distillation type":"ARMADA - Leuclid"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 87.4. Result ID: 69b285886dbb7075b95d6eed. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"BERT-base","Loss":"Leuclid","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 87.4. Result ID: 69b285886dbb7075b95d6f0e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"BERT-base","Distillation type":"ARMADA - Lelementwise"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 85.7. Result ID: 69b285886dbb7075b95d6f0c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"BERT-base","Distillation type":"ARMADA - Lcosine"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 85.7. Result ID: 69b285886dbb7075b95d6f0d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"BERT-base","Distillation type":"ARMADA - Leuclid"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 85.6. Result ID: 69b285886dbb7075b95d6eec. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"BERT-base","Loss":"Lcosine","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 83.8. Result ID: 69b285886dbb7075b95d6eee. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"BERT-base","Loss":"Lelementwise","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### BERT-base CB Accuracy: 83.7. Result ID: 69b285886dbb7075b95d6eeb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Distillation type":"undistilled"}. This record does not assert independent reproduction. ### undistilled CB Accuracy: 83.7. Result ID: 69b285886dbb7075b95d6f0b. [Reporting paper]() Recorded evaluation configuration: {"Model":"BERT-base","Distillation type":"undistilled"}. This record does not assert independent reproduction. ### DeBERTa-v2-xxlarge CB Accuracy: 81.9. Result ID: 69b285886dbb7075b95d6eef. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Distillation type":"undistilled"}. This record does not assert independent reproduction. ### undistilled CB Accuracy: 81.9. Result ID: 69b285886dbb7075b95d6f0f. [Reporting paper]() Recorded evaluation configuration: {"Model":"DeBERTa-v2-xxlarge","Distillation type":"undistilled"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 80.2. Result ID: 69b285886dbb7075b95d6ef4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"OPT-1.3B","Loss":"Lcosine","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 78.6. Result ID: 69b285886dbb7075b95d6f0a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"BERT-6L","Distillation type":"ARMADA - Lelementwise"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 78.5. Result ID: 69b285886dbb7075b95d6ef5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"OPT-1.3B","Loss":"Leuclid","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 78.5. Result ID: 69b285886dbb7075b95d6f09. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"BERT-6L","Distillation type":"ARMADA - Leuclid"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 78.5. Result ID: 69b285886dbb7075b95d6f14. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"OPT-1.3B","Distillation type":"ARMADA - Lcosine"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 78.4. Result ID: 69b285886dbb7075b95d6eea. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"BERT-6L","Loss":"Lelementwise","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### OPT-1.3B CB Accuracy: 78.2. Result ID: 69b285886dbb7075b95d6ef3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Distillation type":"undistilled"}. This record does not assert independent reproduction. ### undistilled CB Accuracy: 78.2. Result ID: 69b285886dbb7075b95d6f13. [Reporting paper]() Recorded evaluation configuration: {"Model":"OPT-1.3B","Distillation type":"undistilled"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 76.8. Result ID: 69b285886dbb7075b95d6f08. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"BERT-6L","Distillation type":"ARMADA - Lcosine"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 76.7. Result ID: 69b285886dbb7075b95d6f15. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"OPT-1.3B","Distillation type":"ARMADA - Leuclid"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 76.7. Result ID: 69b285886dbb7075b95d6f16. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"OPT-1.3B","Distillation type":"ARMADA - Lelementwise"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 76.6. Result ID: 69b285886dbb7075b95d6ef6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"OPT-1.3B","Loss":"Lelementwise","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### BERT-6L CB Accuracy: 75. Result ID: 69b285886dbb7075b95d6ee7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Distillation type":"undistilled"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 75. Result ID: 69b285886dbb7075b95d6ee9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"BERT-6L","Loss":"Leuclid","Teacher":"Midjourney"}. This record does not assert independent reproduction. ### undistilled CB Accuracy: 75. Result ID: 69b285886dbb7075b95d6f07. [Reporting paper]() Recorded evaluation configuration: {"Model":"BERT-6L","Distillation type":"undistilled"}. This record does not assert independent reproduction. ### ARMADA CB Accuracy: 73.1. Result ID: 69b285886dbb7075b95d6ee8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Models":"BERT-6L","Loss":"Lcosine","Teacher":"Midjourney"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.