# Generative Modeling on Fashion MNIST (train) Source: [Generative Modeling on Fashion MNIST (train)]() Collection index: [llms.txt]() Data updated: 2026-03-31T12:22:28.936Z Evaluation of generative models on the training split of the Fashion MNIST dataset, typically assessed using log-likelihood 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 | Generative Modeling | | Dataset (including any stated split) | Fashion MNIST (train) | | Metric | Log Likelihood (100 samples) | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 51 of 51 | | 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 - [Log Likelihood (100 samples)](): Higher is better; unit: SCALAR. - [ELBO (Train)](): Higher is better; unit: SCALAR. - [Log Evidence (Total)](): Higher is better; unit: SCALAR. ## Recorded results | Method | Log Likelihood (100 samples) | Result date | Report paper | | --- | --- | --- | --- | | CARMS-I | -230.77 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-G | -231.16 | Not recorded | 68fa6b171e209131cea2c13c | | LOORF | -231.35 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-G | -231.67 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-I | -231.72 | Not recorded | 68fa6b171e209131cea2c13c | | UNORD | -231.74 | Not recorded | 68fa6b171e209131cea2c13c | | LOORF | -232.01 | Not recorded | 68fa6b171e209131cea2c13c | | UNORD | -232.19 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-I | -233.13 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-G | -233.2 | Not recorded | 68fa6b171e209131cea2c13c | | UNORD | -233.34 | Not recorded | 68fa6b171e209131cea2c13c | | LOORF | -233.75 | Not recorded | 68fa6b171e209131cea2c13c | | ARSM | -235.88 | Not recorded | 68fa6b171e209131cea2c13c | | ARSM | -237.29 | Not recorded | 68fa6b171e209131cea2c13c | | ARSM | -237.93 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-I | -240.44 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-G | -240.52 | Not recorded | 68fa6b171e209131cea2c13c | | LOORF | -240.91 | Not recorded | 68fa6b171e209131cea2c13c | | UNORD | -241.08 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-G | -241.9 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-I | -242.06 | Not recorded | 68fa6b171e209131cea2c13c | | LOORF | -242.17 | Not recorded | 68fa6b171e209131cea2c13c | | UNORD | -242.43 | Not recorded | 68fa6b171e209131cea2c13c | | ARSM | -243.29 | Not recorded | 68fa6b171e209131cea2c13c | | ARSM | -244.63 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-I | -245.44 | Not recorded | 68fa6b171e209131cea2c13c | | CARMS-G | -245.69 | Not recorded | 68fa6b171e209131cea2c13c | | LOORF | -245.8 | Not recorded | 68fa6b171e209131cea2c13c | | UNORD | -245.9 | Not recorded | 68fa6b171e209131cea2c13c | | ARSM | -247.51 | Not recorded | 68fa6b171e209131cea2c13c | ### CARMS-I Log Likelihood (100 samples): -230.77. Result ID: 698e5828ffc3901de09cce9d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"10"}. This record does not assert independent reproduction. ### CARMS-G Log Likelihood (100 samples): -231.16. Result ID: 698e5828ffc3901de09cce9e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"10"}. This record does not assert independent reproduction. ### LOORF Log Likelihood (100 samples): -231.35. Result ID: 698e5828ffc3901de09cce9f. [Reporting paper]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"10"}. This record does not assert independent reproduction. ### CARMS-G Log Likelihood (100 samples): -231.67. Result ID: 698e5828ffc3901de09cce99. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"5"}. This record does not assert independent reproduction. ### CARMS-I Log Likelihood (100 samples): -231.72. Result ID: 698e5828ffc3901de09cce98. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"5"}. This record does not assert independent reproduction. ### UNORD Log Likelihood (100 samples): -231.74. Result ID: 698e5828ffc3901de09ccea0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"10"}. This record does not assert independent reproduction. ### LOORF Log Likelihood (100 samples): -232.01. Result ID: 698e5828ffc3901de09cce9a. [Reporting paper]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"5"}. This record does not assert independent reproduction. ### UNORD Log Likelihood (100 samples): -232.19. Result ID: 698e5828ffc3901de09cce9b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"5"}. This record does not assert independent reproduction. ### CARMS-I Log Likelihood (100 samples): -233.13. Result ID: 698e5828ffc3901de09cce93. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"3"}. This record does not assert independent reproduction. ### CARMS-G Log Likelihood (100 samples): -233.2. Result ID: 698e5828ffc3901de09cce94. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"3"}. This record does not assert independent reproduction. ### UNORD Log Likelihood (100 samples): -233.34. Result ID: 698e5828ffc3901de09cce96. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"3"}. This record does not assert independent reproduction. ### LOORF Log Likelihood (100 samples): -233.75. Result ID: 698e5828ffc3901de09cce95. [Reporting paper]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"3"}. This record does not assert independent reproduction. ### ARSM Log Likelihood (100 samples): -235.88. Result ID: 698e5828ffc3901de09ccea1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"10"}. This record does not assert independent reproduction. ### ARSM Log Likelihood (100 samples): -237.29. Result ID: 698e5828ffc3901de09cce9c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"5"}. This record does not assert independent reproduction. ### ARSM Log Likelihood (100 samples): -237.93. Result ID: 698e5828ffc3901de09cce97. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Nonlinear","Categories":"3"}. This record does not assert independent reproduction. ### CARMS-I Log Likelihood (100 samples): -240.44. Result ID: 698e5828ffc3901de09cce8e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"10"}. This record does not assert independent reproduction. ### CARMS-G Log Likelihood (100 samples): -240.52. Result ID: 698e5828ffc3901de09cce8f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"10"}. This record does not assert independent reproduction. ### LOORF Log Likelihood (100 samples): -240.91. Result ID: 698e5828ffc3901de09cce90. [Reporting paper]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"10"}. This record does not assert independent reproduction. ### UNORD Log Likelihood (100 samples): -241.08. Result ID: 698e5828ffc3901de09cce91. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"10"}. This record does not assert independent reproduction. ### CARMS-G Log Likelihood (100 samples): -241.9. Result ID: 698e5828ffc3901de09cce8a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"5"}. This record does not assert independent reproduction. ### CARMS-I Log Likelihood (100 samples): -242.06. Result ID: 698e5828ffc3901de09cce89. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"5"}. This record does not assert independent reproduction. ### LOORF Log Likelihood (100 samples): -242.17. Result ID: 698e5828ffc3901de09cce8b. [Reporting paper]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"5"}. This record does not assert independent reproduction. ### UNORD Log Likelihood (100 samples): -242.43. Result ID: 698e5828ffc3901de09cce8c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"5"}. This record does not assert independent reproduction. ### ARSM Log Likelihood (100 samples): -243.29. Result ID: 698e5828ffc3901de09cce92. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"10"}. This record does not assert independent reproduction. ### ARSM Log Likelihood (100 samples): -244.63. Result ID: 698e5828ffc3901de09cce8d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"5"}. This record does not assert independent reproduction. ### CARMS-I Log Likelihood (100 samples): -245.44. Result ID: 698e5828ffc3901de09cce84. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"3"}. This record does not assert independent reproduction. ### CARMS-G Log Likelihood (100 samples): -245.69. Result ID: 698e5828ffc3901de09cce85. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"3"}. This record does not assert independent reproduction. ### LOORF Log Likelihood (100 samples): -245.8. Result ID: 698e5828ffc3901de09cce86. [Reporting paper]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"3"}. This record does not assert independent reproduction. ### UNORD Log Likelihood (100 samples): -245.9. Result ID: 698e5828ffc3901de09cce87. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"3"}. This record does not assert independent reproduction. ### ARSM Log Likelihood (100 samples): -247.51. Result ID: 698e5828ffc3901de09cce88. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Network Architecture":"Linear","Categories":"3"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.