File size: 824 Bytes
6706958
06a5dc2
 
6706958
 
 
06a5dc2
6706958
 
726fb9f
 
06a5dc2
965fdb6
06a5dc2
726fb9f
16b3784
 
 
 
 
 
 
 
726fb9f
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
---
title: PhillipsLab
emoji: ⚡️
colorFrom: green
colorTo: blue
sdk: static
pinned: true
---

# PhillipsLab — University of Utah

**Kahlert School of Computing** · Led by [Prof. Jeff M. Phillips](https://users.cs.utah.edu/~jeffp/)

> Interpretable AI · Activation Steering · Dataset Curation  

---


## 📊 Featured Datasets

- [Conciseness ↔ Verbosity Contrast](https://huggingface.co/datasets/PhillipsLab/conciseness_verbosity_contrast) — minimal pairs isolating **conciseness vs verbosity** for interpretability and activation steering.  
- [Positivity ↔ Negativity Contrast](https://huggingface.co/datasets/PhillipsLab/positivity_negativity_contrast) — stance-style pairs controlling **positive vs negative sentiment** while keeping factual parity.


---

*PhillipsLab © 2025 — University of Utah*