# Bias Mitigation for Stereotype Queries on UTKFACE Gender Source: [Bias Mitigation for Stereotype Queries on UTKFACE Gender]() Collection index: [llms.txt]() Data updated: 2026-03-20T10:21:39.813Z Evaluation of fairness and bias mitigation techniques using KL Divergence and MaxSkew metrics on the UTKFACE dataset, focusing on gender stereotypes. 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 | Bias Mitigation for Stereotype Queries | | Dataset (including any stated split) | UTKFACE Gender | | Metric | KL Divergence | | Unit | SCALAR | | Direction | Lower is better | | Results loaded | 33 of 33 | | 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 - [KL Divergence](): Lower is better; unit: SCALAR. - [MaxSkew](): Lower is better; unit: SCALAR. ## Recorded results | Method | KL Divergence | Result date | Report paper | | --- | --- | --- | --- | | BEND-VLM | 0.004 | Not recorded | 68fa759e1e209131cea68fe9 | | BEND-VLM | 0.008 | Not recorded | 68fa759e1e209131cea68fe9 | | BENDVLM | 0.008 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | BENDSEMbi | 0.009 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | BENDSEMbi | 0.012 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | BENDVLM | 0.013 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | ZSDEBIAS | 0.022 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | Baseline CLIP | 0.029 | Not recorded | 68fa759e1e209131cea68fe9 | | Orth-Cal. | 0.03 | Not recorded | 68fa759e1e209131cea68fe9 | | ZSDEBIAS | 0.036 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | SEMi | 0.041 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | Orth-Proj. | 0.043 | Not recorded | 68fa759e1e209131cea68fe9 | | SEMbi | 0.069 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | SEMb | 0.07 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | SEMb | 0.077 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | SEMbi | 0.081 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | ORTH-CALI | 0.086 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | PRISM | 0.088 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | DebiasCLIP | 0.091 | Not recorded | 68fa759e1e209131cea68fe9 | | SEMi | 0.111 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | ORTH-PROJ | 0.117 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | PRISM-MINI | 0.117 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | Baseline CLIP | 0.12 | Not recorded | 68fa759e1e209131cea68fe9 | | PRISM | 0.127 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | BASE CLIP | 0.136 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | BASE CLIP | 0.153 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | ORTH-PROJ | 0.163 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | PRISM-MINI | 0.168 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | Orth-Proj. | 0.191 | Not recorded | 68fa759e1e209131cea68fe9 | | ORTH-CALI | 0.204 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | Orth-Cal. | 0.254 | Not recorded | 68fa759e1e209131cea68fe9 | | ROBOSHOT | 0.258 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | | ROBOSHOT | 0.323 | Not recorded | 69bcd4ab0636e5bf4afc2bf4 | ### BEND-VLM KL Divergence: 0.004. Result ID: 6993f3423a6c31484919b903. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"CLIP-VIT-L-P14"}. This record does not assert independent reproduction. ### BEND-VLM KL Divergence: 0.008. Result ID: 6993f3423a6c31484919b8ff. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"CLIP-VIT-B-P16"}. This record does not assert independent reproduction. ### BENDVLM KL Divergence: 0.008. Result ID: 69bd1f9647f34ce924a7e1c7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"false","Prompting Category":"Bias prompts + input-specific prompts + labeled images"}. This record does not assert independent reproduction. ### BENDSEMbi KL Divergence: 0.009. Result ID: 69bd1f9647f34ce924a7e1c9. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"false","Prompting Category":"Bias prompts + input-specific prompts + labeled images"}. This record does not assert independent reproduction. ### BENDSEMbi KL Divergence: 0.012. Result ID: 69bd1f9647f34ce924a7e1ca. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"false","Prompting Category":"Bias prompts + input-specific prompts + labeled images"}. This record does not assert independent reproduction. ### BENDVLM KL Divergence: 0.013. Result ID: 69bd1f9647f34ce924a7e1c8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"false","Prompting Category":"Bias prompts + input-specific prompts + labeled images"}. This record does not assert independent reproduction. ### ZSDEBIAS KL Divergence: 0.022. Result ID: 69bd1f9647f34ce924a7e1c0. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true","Prompting Category":"Bias prompts only"}. This record does not assert independent reproduction. ### Baseline CLIP KL Divergence: 0.029. Result ID: 6993f3423a6c31484919b900. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"CLIP-VIT-L-P14"}. This record does not assert independent reproduction. ### Orth-Cal. KL Divergence: 0.03. Result ID: 6993f3423a6c31484919b902. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"CLIP-VIT-L-P14"}. This record does not assert independent reproduction. ### ZSDEBIAS KL Divergence: 0.036. Result ID: 69bd1f9647f34ce924a7e1bf. [Reporting paper]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true","Prompting Category":"Bias prompts only"}. This record does not assert independent reproduction. ### SEMi KL Divergence: 0.041. Result ID: 69bd1f9647f34ce924a7e1b8. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true","Prompting Category":"Bias-agnostic + input-specific prompts"}. This record does not assert independent reproduction. ### Orth-Proj. KL Divergence: 0.043. Result ID: 6993f3423a6c31484919b901. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"CLIP-VIT-L-P14"}. This record does not assert independent reproduction. ### SEMbi KL Divergence: 0.069. Result ID: 69bd1f9647f34ce924a7e1c4. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true","Prompting Category":"Bias prompts + input-specific prompts"}. This record does not assert independent reproduction. ### SEMb KL Divergence: 0.07. Result ID: 69bd1f9647f34ce924a7e1be. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true","Prompting Category":"Bias prompts only"}. This record does not assert independent reproduction. ### SEMb KL Divergence: 0.077. Result ID: 69bd1f9647f34ce924a7e1bd. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true","Prompting Category":"Bias prompts only"}. This record does not assert independent reproduction. ### SEMbi KL Divergence: 0.081. Result ID: 69bd1f9647f34ce924a7e1c3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true","Prompting Category":"Bias prompts + input-specific prompts"}. This record does not assert independent reproduction. ### ORTH-CALI KL Divergence: 0.086. Result ID: 69bd1f9647f34ce924a7e1c1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true","Prompting Category":"Bias prompts + input-specific prompts"}. This record does not assert independent reproduction. ### PRISM KL Divergence: 0.088. Result ID: 69bd1f9647f34ce924a7e1c5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true","Prompting Category":"Bias prompts + input-specific prompts"}. This record does not assert independent reproduction. ### DebiasCLIP KL Divergence: 0.091. Result ID: 6993f3423a6c31484919b8fe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"CLIP-VIT-B-P16"}. This record does not assert independent reproduction. ### SEMi KL Divergence: 0.111. Result ID: 69bd1f9647f34ce924a7e1b7. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true","Prompting Category":"Bias-agnostic + input-specific prompts"}. This record does not assert independent reproduction. ### ORTH-PROJ KL Divergence: 0.117. Result ID: 69bd1f9647f34ce924a7e1b9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true","Prompting Category":"Bias prompts only"}. This record does not assert independent reproduction. ### PRISM-MINI KL Divergence: 0.117. Result ID: 69bd1f9647f34ce924a7e1bb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true","Prompting Category":"Bias prompts only"}. This record does not assert independent reproduction. ### Baseline CLIP KL Divergence: 0.12. Result ID: 6993f3423a6c31484919b8fb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"CLIP-VIT-B-P16"}. This record does not assert independent reproduction. ### PRISM KL Divergence: 0.127. Result ID: 69bd1f9647f34ce924a7e1c6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true","Prompting Category":"Bias prompts + input-specific prompts"}. This record does not assert independent reproduction. ### BASE CLIP KL Divergence: 0.136. Result ID: 69bd1f9647f34ce924a7e1b4. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true"}. This record does not assert independent reproduction. ### BASE CLIP KL Divergence: 0.153. Result ID: 69bd1f9647f34ce924a7e1b3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true"}. This record does not assert independent reproduction. ### ORTH-PROJ KL Divergence: 0.163. Result ID: 69bd1f9647f34ce924a7e1ba. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true","Prompting Category":"Bias prompts only"}. This record does not assert independent reproduction. ### PRISM-MINI KL Divergence: 0.168. Result ID: 69bd1f9647f34ce924a7e1bc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true","Prompting Category":"Bias prompts only"}. This record does not assert independent reproduction. ### Orth-Proj. KL Divergence: 0.191. Result ID: 6993f3423a6c31484919b8fc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"CLIP-VIT-B-P16"}. This record does not assert independent reproduction. ### ORTH-CALI KL Divergence: 0.204. Result ID: 69bd1f9647f34ce924a7e1c2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true","Prompting Category":"Bias prompts + input-specific prompts"}. This record does not assert independent reproduction. ### Orth-Cal. KL Divergence: 0.254. Result ID: 6993f3423a6c31484919b8fd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"CLIP-VIT-B-P16"}. This record does not assert independent reproduction. ### ROBOSHOT KL Divergence: 0.258. Result ID: 69bd1f9647f34ce924a7e1b5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-50","Zero-shot":"true","Prompting Category":"Bias-agnostic + input-specific prompts"}. This record does not assert independent reproduction. ### ROBOSHOT KL Divergence: 0.323. Result ID: 69bd1f9647f34ce924a7e1b6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"ResNet-101","Zero-shot":"true","Prompting Category":"Bias-agnostic + input-specific prompts"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.