DontPlanToEnd
commited on
Commit
•
005ff12
1
Parent(s):
cb863bc
Update app.py
Browse files
app.py
CHANGED
@@ -56,19 +56,19 @@ GraInter = gr.Blocks()
|
|
56 |
with GraInter:
|
57 |
gr.HTML("""<h1 align="center">UGI Leaderboard</h1>""")
|
58 |
gr.Markdown("""
|
59 |
-
UGI: Uncensored General Intelligence
|
60 |
|
61 |
-
W/10
|
62 |
|
63 |
-
Unruly
|
64 |
|
65 |
-
Internet
|
66 |
|
67 |
-
CrimeStats
|
68 |
|
69 |
-
Stories/Jokes
|
70 |
|
71 |
-
PolContro
|
72 |
""")
|
73 |
with gr.Column():
|
74 |
with gr.Row():
|
|
|
56 |
with GraInter:
|
57 |
gr.HTML("""<h1 align="center">UGI Leaderboard</h1>""")
|
58 |
gr.Markdown("""
|
59 |
+
**UGI: Uncensored General Intelligence**. The average of 5 different subjects that LLMs are commonly steered away from. The leaderboard is made of roughly 60 questions/tasks, measuring both "willingness to answer" and "accuracy" in fact-based controversial questions.
|
60 |
|
61 |
+
**W/10:** A more narrow, 10-point score, solely measuring the LLM's Willingness to answer controversial questions.
|
62 |
|
63 |
+
**Unruly:** Knowledge of activities that are generally frowned upon.
|
64 |
|
65 |
+
**Internet:** Knowledge of various internet information, from professional to deviant.
|
66 |
|
67 |
+
**CrimeStats:** Knowledge of crime statistics which are uncomfortable to talk about.
|
68 |
|
69 |
+
**Stories/Jokes:** Ability to write offensive stories and jokes.
|
70 |
|
71 |
+
**PolContro:** Knowledge of politically/socially controversial information.
|
72 |
""")
|
73 |
with gr.Column():
|
74 |
with gr.Row():
|