# Natural Language Inference on SciTail Source: [Natural Language Inference on SciTail]() Collection index: [llms.txt]() Data updated: 2026-05-29T10:38:10.485Z Evaluation of Natural Language Inference capabilities using the SciTail dataset, typically measured by Accuracy. 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 Inference | | Dataset (including any stated split) | SciTail | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 39 of 39 | | 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 - [Accuracy](): Higher is better; unit: PERCENT. - [Accuracy (Normalized)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Individual Task | 96.52 | Not recorded | 69e705037d0172140ec7f1b2 | | CoreSpace + ENMP | 94.97 | Not recorded | 69e705037d0172140ec7f1b2 | | Finetuned | 94.8 | Not recorded | 69ca036e0636e5bf4afc9292 | | CoreSpace | 94.45 | Not recorded | 69e705037d0172140ec7f1b2 | | TSV + ENMP | 94.18 | Not recorded | 69e705037d0172140ec7f1b2 | | TSV | 94.17 | Not recorded | 69e705037d0172140ec7f1b2 | | TA + ENMP | 93.98 | Not recorded | 69e705037d0172140ec7f1b2 | | KnOTS | 93.98 | Not recorded | 69e705037d0172140ec7f1b2 | | KnOTS + ENMP | 93.85 | Not recorded | 69e705037d0172140ec7f1b2 | | DARE + ENMP | 93.84 | Not recorded | 69e705037d0172140ec7f1b2 | | TA | 93.32 | Not recorded | 69e705037d0172140ec7f1b2 | | TIES + ENMP | 93.04 | Not recorded | 69e705037d0172140ec7f1b2 | | DARE | 92.8 | Not recorded | 69e705037d0172140ec7f1b2 | | TIES | 92.66 | Not recorded | 69e705037d0172140ec7f1b2 | | DPS Mix | 80.36 | Not recorded | 68fa6ddd1e209131cea3c871 | | CHILD-TUNINGD | 79.86 | Not recorded | 68fa6ddd1e209131cea3c871 | | vanilla | 79.7 | Not recorded | 68fa6ddd1e209131cea3c871 | | DPS Dense | 79.35 | Not recorded | 68fa6ddd1e209131cea3c871 | | DPS Dense | 78.42 | Not recorded | 68fa6ddd1e209131cea3c871 | | vanilla | 78.22 | Not recorded | 68fa6ddd1e209131cea3c871 | | DPS Mix | 77.8 | Not recorded | 68fa6ddd1e209131cea3c871 | | CHILD-TUNINGD | 77.17 | Not recorded | 68fa6ddd1e209131cea3c871 | | MIC | 74.5 | Not recorded | 6a191ddedbd7150506f82b9b | | MIC | 74.18 | Not recorded | 6a191ddedbd7150506f82b9b | | MIC | 73.09 | Not recorded | 6a191ddedbd7150506f82b9b | | MIC | 71.78 | Not recorded | 6a191ddedbd7150506f82b9b | ### Individual Task Accuracy: 96.52. Result ID: 69e7645247f34ce924cec943. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Evaluation Protocol":"fine-tuned on single tasks"}. This record does not assert independent reproduction. ### CoreSpace + ENMP Accuracy: 94.97. Result ID: 69e7645247f34ce924cec94f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"0.5","ENMP (Elective Neuron Modification Protocol)":"true"}. This record does not assert independent reproduction. ### Finetuned Accuracy: 94.8. Result ID: 69ca276c47f34ce924b27650. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"LLAMA-3 8B","LoRA rank":"16"}. This record does not assert independent reproduction. ### CoreSpace Accuracy: 94.45. Result ID: 69e7645247f34ce924cec94e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"0.5"}. This record does not assert independent reproduction. ### TSV + ENMP Accuracy: 94.18. Result ID: 69e7645247f34ce924cec94b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"0.6","ENMP (Elective Neuron Modification Protocol)":"true"}. This record does not assert independent reproduction. ### TSV Accuracy: 94.17. Result ID: 69e7645247f34ce924cec94a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"0.6"}. This record does not assert independent reproduction. ### TA + ENMP Accuracy: 93.98. Result ID: 69e7645247f34ce924cec945. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"0.3","ENMP (Elective Neuron Modification Protocol)":"true"}. This record does not assert independent reproduction. ### KnOTS Accuracy: 93.98. Result ID: 69e7645247f34ce924cec94c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"1.1","Top-k (%)":"90"}. This record does not assert independent reproduction. ### KnOTS + ENMP Accuracy: 93.85. Result ID: 69e7645247f34ce924cec94d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"1.1","Top-k (%)":"90","ENMP (Elective Neuron Modification Protocol)":"true"}. This record does not assert independent reproduction. ### DARE + ENMP Accuracy: 93.84. Result ID: 69e7645247f34ce924cec949. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"1.1","Top-k (%)":"80","Drop rate (p)":"0.1","ENMP (Elective Neuron Modification Protocol)":"true"}. This record does not assert independent reproduction. ### TA Accuracy: 93.32. Result ID: 69e7645247f34ce924cec944. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"0.3"}. This record does not assert independent reproduction. ### TIES + ENMP Accuracy: 93.04. Result ID: 69e7645247f34ce924cec947. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"1.2","Top-k (%)":"80","ENMP (Elective Neuron Modification Protocol)":"true"}. This record does not assert independent reproduction. ### DARE Accuracy: 92.8. Result ID: 69e7645247f34ce924cec948. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"1.1","Top-k (%)":"80","Drop rate (p)":"0.1"}. This record does not assert independent reproduction. ### TIES Accuracy: 92.66. Result ID: 69e7645247f34ce924cec946. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Llama-3","Scaling factor (λ)":"1.2","Top-k (%)":"80"}. This record does not assert independent reproduction. ### DPS Mix Accuracy: 80.36. Result ID: 698fe1163a6c314849fc8255. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"MNLI"}. This record does not assert independent reproduction. ### CHILD-TUNINGD Accuracy: 79.86. Result ID: 698fe1163a6c314849fc8253. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Dataset":"MNLI"}. This record does not assert independent reproduction. ### vanilla Accuracy: 79.7. Result ID: 698fe1163a6c314849fc8252. [Reporting paper]() Recorded evaluation configuration: {"Training Dataset":"MNLI"}. This record does not assert independent reproduction. ### DPS Dense Accuracy: 79.35. Result ID: 698fe1163a6c314849fc8254. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"MNLI"}. This record does not assert independent reproduction. ### DPS Dense Accuracy: 78.42. Result ID: 698fe1163a6c314849fc8258. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"SNLI"}. This record does not assert independent reproduction. ### vanilla Accuracy: 78.22. Result ID: 698fe1163a6c314849fc8256. [Reporting paper]() Recorded evaluation configuration: {"Training Dataset":"SNLI"}. This record does not assert independent reproduction. ### DPS Mix Accuracy: 77.8. Result ID: 698fe1163a6c314849fc8259. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Training Dataset":"SNLI"}. This record does not assert independent reproduction. ### CHILD-TUNINGD Accuracy: 77.17. Result ID: 698fe1163a6c314849fc8257. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Training Dataset":"SNLI"}. This record does not assert independent reproduction. ### MIC Accuracy: 74.5. Result ID: 6a196c8cfbc373592c14754e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"TinyBERT 6L","Dimension":"768"}. This record does not assert independent reproduction. ### MIC Accuracy: 74.18. Result ID: 6a196c8cfbc373592c14754f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"BERT","Dimension":"768"}. This record does not assert independent reproduction. ### MIC Accuracy: 73.09. Result ID: 6a196c8cfbc373592c14754d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"BERT","Dimension":"16"}. This record does not assert independent reproduction. ### MIC Accuracy: 71.78. Result ID: 6a196c8cfbc373592c14754c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"TinyBERT 6L","Dimension":"16"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.