Datasets:
ArXiv:
DOI:
License:
Commit
•
3b9f78b
1
Parent(s):
bb40e2a
random commit!
Browse files
README.md
CHANGED
|
@@ -207,7 +207,7 @@ Documentation is a great first step for sharing general insights about a model
|
|
| 207 |
</p>
|
| 208 |
|
| 209 |
Visualization of model outputs isn’t just for generative models though! For classification models, we also want to look out for bias-related harms caused by a model’s **disparate performance** on different demographics. If you know what protected classes are most at risk of discrimination and have those annotated in an evaluation set, then you can report disaggregated performance over the different categories in [your model card](https://dl.acm.org/doi/10.1145/3287560.3287596) as mentioned above, so users can make informed decisions. If however, you are worried that you haven’t identified all populations at risk of bias-related harms, or if you do not have access to annotated test examples to measure the biases you suspect, that’s where interactive visualizations of where and how the model fails come in handy! To help you with this, the [SEAL app](https://huggingface.co/spaces/nazneen/seal) groups similar mistakes by your model and shows you some common features in each cluster. If you want to go further, you can even combine it with the [disaggregators library](https://github.com/huggingface/disaggregators) we introduced in the datasets section to find clusters that are indicative of bias-related failure modes!
|
| 210 |
-
|
| 211 |
|
| 212 |
<p align="center">
|
| 213 |
<br>
|
|
|
|
| 207 |
</p>
|
| 208 |
|
| 209 |
Visualization of model outputs isn’t just for generative models though! For classification models, we also want to look out for bias-related harms caused by a model’s **disparate performance** on different demographics. If you know what protected classes are most at risk of discrimination and have those annotated in an evaluation set, then you can report disaggregated performance over the different categories in [your model card](https://dl.acm.org/doi/10.1145/3287560.3287596) as mentioned above, so users can make informed decisions. If however, you are worried that you haven’t identified all populations at risk of bias-related harms, or if you do not have access to annotated test examples to measure the biases you suspect, that’s where interactive visualizations of where and how the model fails come in handy! To help you with this, the [SEAL app](https://huggingface.co/spaces/nazneen/seal) groups similar mistakes by your model and shows you some common features in each cluster. If you want to go further, you can even combine it with the [disaggregators library](https://github.com/huggingface/disaggregators) we introduced in the datasets section to find clusters that are indicative of bias-related failure modes!
|
| 210 |
+
|
| 211 |
|
| 212 |
<p align="center">
|
| 213 |
<br>
|