Spaces:
Runtime error
Runtime error
Layout preparation for prompt mode and chat mode.
Browse files
app.py
ADDED
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
|
3 |
+
from prompt import iface_prompt
|
4 |
+
|
5 |
+
with gr.Blocks() as iface:
|
6 |
+
gr.Markdown("""# Petals playground
|
7 |
+
**Let's play with prompts and inference settings for BLOOM and BLOOMZ 176B models!**
|
8 |
+
|
9 |
+
This space uses websocket API of [chat.petals.ml](http://chat.petals.ml). Health status of Petals network [lives here](http://health.petals.ml).
|
10 |
+
|
11 |
+
Do NOT talk to BLOOM as an entity, it's not a chatbot but a webpage/blog/article completion model.
|
12 |
+
For the best results: MIMIC a few sentences of a webpage similar to the content you want to generate.
|
13 |
+
|
14 |
+
BLOOMZ performs better in chat mode and understands the instructions better.""")
|
15 |
+
|
16 |
+
gr.TabbedInterface([iface_prompt, ], ["Prompt mode",])
|
17 |
+
|
18 |
+
# Queues are required to enable generators
|
19 |
+
iface.queue(concurrency_count=5)
|
20 |
+
iface.launch()
|
prompt.py
CHANGED
@@ -70,16 +70,8 @@ def generate(prompt, model, endseq, max_length,
|
|
70 |
yield [prompt, output + "\nError: " + traceback.format_exc()]
|
71 |
return
|
72 |
|
73 |
-
with gr.Blocks() as
|
74 |
-
gr.Markdown("""
|
75 |
-
**Let's play with prompts and inference settings for BLOOM and BLOOMZ 176B models!**
|
76 |
-
|
77 |
-
This space uses websocket API of [chat.petals.ml](http://chat.petals.ml). Health status of Petals network [lives here](http://health.petals.ml).
|
78 |
-
|
79 |
-
Do NOT talk to BLOOM as an entity, it's not a chatbot but a webpage/blog/article completion model.
|
80 |
-
For the best results: MIMIC a few sentences of a webpage similar to the content you want to generate.
|
81 |
-
|
82 |
-
BLOOMZ performs better in chat mode and understands the instructions better.""")
|
83 |
|
84 |
with gr.Row():
|
85 |
model = gr.Radio(["bloom", "bloomz", "bloom-7b1"], value='bloom', label="Use model")
|
@@ -131,7 +123,3 @@ with gr.Blocks() as iface:
|
|
131 |
"Human: What's the capital of Portugal?</s>\n"
|
132 |
"AI: ", "bloomz", True, 0, 0.9, 0.75, False]
|
133 |
])
|
134 |
-
|
135 |
-
# Queues are required to enable generators
|
136 |
-
iface.queue(concurrency_count=5)
|
137 |
-
iface.launch()
|
|
|
70 |
yield [prompt, output + "\nError: " + traceback.format_exc()]
|
71 |
return
|
72 |
|
73 |
+
with gr.Blocks() as iface_prompt:
|
74 |
+
gr.Markdown("""**Useful for testing raw prompts with zero, one or few-shot prompting.**""")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
75 |
|
76 |
with gr.Row():
|
77 |
model = gr.Radio(["bloom", "bloomz", "bloom-7b1"], value='bloom', label="Use model")
|
|
|
123 |
"Human: What's the capital of Portugal?</s>\n"
|
124 |
"AI: ", "bloomz", True, 0, 0.9, 0.75, False]
|
125 |
])
|
|
|
|
|
|
|
|