Update app.py
Browse files
app.py
CHANGED
|
@@ -43,10 +43,10 @@ interface = gr.Interface(
|
|
| 43 |
outputs=gr.Textbox(label="Model Response"),
|
| 44 |
title="Hello, I'm Aligned AI!",
|
| 45 |
description=(
|
| 46 |
-
"""
|
| 47 |
-
This is RLLMv1, the first RLLM prototype that took a staggering 24 layers of sequential training.
|
| 48 |
-
The main issues with this model
|
| 49 |
-
You can read my rough post on this model <a href=https://www.lesswrong.com/posts/GrxaMeekGKK6WKwmm/rl-for-safety-work-or-just-clever-rl-reinforcement-learning >here</a>.
|
| 50 |
"""
|
| 51 |
),
|
| 52 |
)
|
|
|
|
| 43 |
outputs=gr.Textbox(label="Model Response"),
|
| 44 |
title="Hello, I'm Aligned AI!",
|
| 45 |
description=(
|
| 46 |
+
"""
|
| 47 |
+
This is RLLMv1, the first RLLM prototype that took a staggering 24 layers of sequential training.
|
| 48 |
+
The main issues with this model are that it is slow and is too preoccupied with ethical alignment.
|
| 49 |
+
You can read my rough post on this model <a href=https://www.lesswrong.com/posts/GrxaMeekGKK6WKwmm/rl-for-safety-work-or-just-clever-rl-reinforcement-learning >here</a>.
|
| 50 |
"""
|
| 51 |
),
|
| 52 |
)
|