# Few-shot classification on Omniglot 5-way 5-shot (test) Source: [Few-shot classification on Omniglot 5-way 5-shot (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:28:12.406Z Evaluation of few-shot classification performance on the test split of the Omniglot dataset using the 5-way, 5-shot configuration, 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 | Few-shot classification | | Dataset (including any stated split) | Omniglot 5-way 5-shot (test) | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 35 of 35 | | 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. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | ProtoNets | 99.58 | Not recorded | 68fa684f1e209131cea1b543 | | Units-ML | 99.23 | Not recorded | 68fa6c0f1e209131cea31d3f | | Units-ML | 99.16 | Not recorded | 68fa6c0f1e209131cea31d3f | | Units-ML | 98.86 | Not recorded | 68fa6c0f1e209131cea31d3f | | MAML | 98.83 | Not recorded | 68fa684f1e209131cea1b543 | | Units-NTS | 98.46 | Not recorded | 68fa6c0f1e209131cea31d3f | | Units-NTS | 98.43 | Not recorded | 68fa6c0f1e209131cea31d3f | | ULDA-ProtoNets | 98.14 | Not recorded | 68fa684f1e209131cea1b543 | | Units-NTS | 98.13 | Not recorded | 68fa6c0f1e209131cea31d3f | | AAL-MAML++ | 97.96 | Not recorded | 68fa684f1e209131cea1b543 | | ULDA-MetaOptNet | 97.6 | Not recorded | 68fa684f1e209131cea1b543 | | UA-MetaSGD TS | 97.48 | Not recorded | 68fa6c0f1e209131cea31d3f | | UA-MetaSGD TS | 97 | Not recorded | 68fa6c0f1e209131cea31d3f | | MAML | 96.38 | Not recorded | 68fa6c0f1e209131cea31d3f | | MAML | 96.26 | Not recorded | 68fa6c0f1e209131cea31d3f | | UMTRA | 95.43 | Not recorded | 68fa684f1e209131cea1b543 | | UA-MetaSGD NTS | 95.38 | Not recorded | 68fa6c0f1e209131cea31d3f | | UA-MetaSGD NTS | 94.96 | Not recorded | 68fa6c0f1e209131cea31d3f | | UA-MetaSGD TS | 94.32 | Not recorded | 68fa6c0f1e209131cea31d3f | | Versa NTS | 94.03 | Not recorded | 68fa6c0f1e209131cea31d3f | | Versa TS | 93.6 | Not recorded | 68fa6c0f1e209131cea31d3f | | Versa TS | 93.3 | Not recorded | 68fa6c0f1e209131cea31d3f | | UA-MetaSGD NTS | 93.28 | Not recorded | 68fa6c0f1e209131cea31d3f | | MAML | 92 | Not recorded | 68fa6c0f1e209131cea31d3f | | Versa NTS | 91.93 | Not recorded | 68fa6c0f1e209131cea31d3f | | AAL-ProtoNets | 88.41 | Not recorded | 68fa684f1e209131cea1b543 | | CACTUS-MAML | 87.78 | Not recorded | 68fa684f1e209131cea1b543 | | Versa TS | 84.97 | Not recorded | 68fa6c0f1e209131cea31d3f | | Versa NTS | 83.97 | Not recorded | 68fa6c0f1e209131cea31d3f | | CACTUS-ProtoNets | 83.58 | Not recorded | 68fa684f1e209131cea1b543 | | Training from scratch | 74.78 | Not recorded | 68fa684f1e209131cea1b543 | | linear classifier | 68.72 | Not recorded | 68fa684f1e209131cea1b543 | | knn-nearest neighbors | 68.06 | Not recorded | 68fa684f1e209131cea1b543 | | MLP with dropout | 62.56 | Not recorded | 68fa684f1e209131cea1b543 | | cluster matching | 58.62 | Not recorded | 68fa684f1e209131cea1b543 | ### ProtoNets Accuracy: 99.58. Result ID: 698d698d8b517eb433214cc4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Clustering":"N/A","Supervised":true}. This record does not assert independent reproduction. ### Units-ML Accuracy: 99.23. Result ID: 698ec33cffc3901de0a0e597. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"8000 Tasks","Task selection protocol":"ML"}. This record does not assert independent reproduction. ### Units-ML Accuracy: 99.16. Result ID: 698ec33cffc3901de0a0e598. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"10000 Tasks","Task selection protocol":"ML"}. This record does not assert independent reproduction. ### Units-ML Accuracy: 98.86. Result ID: 698ec33cffc3901de0a0e596. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"4000 Tasks","Task selection protocol":"ML"}. This record does not assert independent reproduction. ### MAML Accuracy: 98.83. Result ID: 698d698d8b517eb433214cc5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Clustering":"N/A","Supervised":true}. This record does not assert independent reproduction. ### Units-NTS Accuracy: 98.46. Result ID: 698ec33cffc3901de0a0e595. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"10000 Tasks","Task selection protocol":"NTS"}. This record does not assert independent reproduction. ### Units-NTS Accuracy: 98.43. Result ID: 698ec33cffc3901de0a0e594. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"8000 Tasks","Task selection protocol":"NTS"}. This record does not assert independent reproduction. ### ULDA-ProtoNets Accuracy: 98.14. Result ID: 698d698d8b517eb433214cc2. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Clustering":"N/A"}. This record does not assert independent reproduction. ### Units-NTS Accuracy: 98.13. Result ID: 698ec33cffc3901de0a0e593. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"4000 Tasks","Task selection protocol":"NTS"}. This record does not assert independent reproduction. ### AAL-MAML++ Accuracy: 97.96. Result ID: 698d698d8b517eb433214cbe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Clustering":"N/A"}. This record does not assert independent reproduction. ### ULDA-MetaOptNet Accuracy: 97.6. Result ID: 698d698d8b517eb433214cc3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Clustering":"N/A"}. This record does not assert independent reproduction. ### UA-MetaSGD TS Accuracy: 97.48. Result ID: 698ec33cffc3901de0a0e592. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"10000 Tasks","Task selection protocol":"TS"}. This record does not assert independent reproduction. ### UA-MetaSGD TS Accuracy: 97. Result ID: 698ec33cffc3901de0a0e591. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"8000 Tasks","Task selection protocol":"TS"}. This record does not assert independent reproduction. ### MAML Accuracy: 96.38. Result ID: 698ec33cffc3901de0a0e586. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"10000 Tasks","Task selection protocol":"Random"}. This record does not assert independent reproduction. ### MAML Accuracy: 96.26. Result ID: 698ec33cffc3901de0a0e585. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"8000 Tasks","Task selection protocol":"Random"}. This record does not assert independent reproduction. ### UMTRA Accuracy: 95.43. Result ID: 698d698d8b517eb433214cc1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Clustering":"N/A"}. This record does not assert independent reproduction. ### UA-MetaSGD NTS Accuracy: 95.38. Result ID: 698ec33cffc3901de0a0e58f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"10000 Tasks","Task selection protocol":"NTS"}. This record does not assert independent reproduction. ### UA-MetaSGD NTS Accuracy: 94.96. Result ID: 698ec33cffc3901de0a0e58e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"8000 Tasks","Task selection protocol":"NTS"}. This record does not assert independent reproduction. ### UA-MetaSGD TS Accuracy: 94.32. Result ID: 698ec33cffc3901de0a0e590. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"4000 Tasks","Task selection protocol":"TS"}. This record does not assert independent reproduction. ### Versa NTS Accuracy: 94.03. Result ID: 698ec33cffc3901de0a0e589. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"10000 Tasks","Task selection protocol":"NTS"}. This record does not assert independent reproduction. ### Versa TS Accuracy: 93.6. Result ID: 698ec33cffc3901de0a0e58c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"10000 Tasks","Task selection protocol":"TS"}. This record does not assert independent reproduction. ### Versa TS Accuracy: 93.3. Result ID: 698ec33cffc3901de0a0e58b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"8000 Tasks","Task selection protocol":"TS"}. This record does not assert independent reproduction. ### UA-MetaSGD NTS Accuracy: 93.28. Result ID: 698ec33cffc3901de0a0e58d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"4000 Tasks","Task selection protocol":"NTS"}. This record does not assert independent reproduction. ### MAML Accuracy: 92. Result ID: 698ec33cffc3901de0a0e584. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"4000 Tasks","Task selection protocol":"Random"}. This record does not assert independent reproduction. ### Versa NTS Accuracy: 91.93. Result ID: 698ec33cffc3901de0a0e588. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"8000 Tasks","Task selection protocol":"NTS"}. This record does not assert independent reproduction. ### AAL-ProtoNets Accuracy: 88.41. Result ID: 698d698d8b517eb433214cbd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Clustering":"N/A"}. This record does not assert independent reproduction. ### CACTUS-MAML Accuracy: 87.78. Result ID: 698d698d8b517eb433214cc0. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Clustering":"ACAI"}. This record does not assert independent reproduction. ### Versa TS Accuracy: 84.97. Result ID: 698ec33cffc3901de0a0e58a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"4000 Tasks","Task selection protocol":"TS"}. This record does not assert independent reproduction. ### Versa NTS Accuracy: 83.97. Result ID: 698ec33cffc3901de0a0e587. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Labeling budget (Number of Tasks)":"4000 Tasks","Task selection protocol":"NTS"}. This record does not assert independent reproduction. ### CACTUS-ProtoNets Accuracy: 83.58. Result ID: 698d698d8b517eb433214cbf. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Clustering":"ACAI"}. This record does not assert independent reproduction. ### Training from scratch Accuracy: 74.78. Result ID: 698d698d8b517eb433214cb8. [Reporting paper]() Recorded evaluation configuration: {"Clustering":"N/A"}. This record does not assert independent reproduction. ### linear classifier Accuracy: 68.72. Result ID: 698d698d8b517eb433214cba. [Reporting paper]() Recorded evaluation configuration: {"Clustering":"DeepCluster"}. This record does not assert independent reproduction. ### knn-nearest neighbors Accuracy: 68.06. Result ID: 698d698d8b517eb433214cb9. [Reporting paper]() Recorded evaluation configuration: {"Clustering":"DeepCluster"}. This record does not assert independent reproduction. ### MLP with dropout Accuracy: 62.56. Result ID: 698d698d8b517eb433214cbb. [Reporting paper]() Recorded evaluation configuration: {"Clustering":"DeepCluster"}. This record does not assert independent reproduction. ### cluster matching Accuracy: 58.62. Result ID: 698d698d8b517eb433214cbc. [Reporting paper]() Recorded evaluation configuration: {"Clustering":"DeepCluster"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.