Spaces:
Runtime error
Runtime error
Create app.py
Browse filesupdates iya yin
app.py
ADDED
@@ -0,0 +1,91 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from gradio_client import Client
|
3 |
+
|
4 |
+
#fusecap_client = Client("https://noamrot-fusecap-image-captioning.hf.space/")
|
5 |
+
fuyu_client = Client("https://adept-fuyu-8b-demo.hf.space/")
|
6 |
+
|
7 |
+
def get_caption(image_in):
|
8 |
+
|
9 |
+
fuyu_result = fuyu_client.predict(
|
10 |
+
image_in, # str representing input in 'raw_image' Image component
|
11 |
+
True, # bool in 'Enable detailed captioning' Checkbox component
|
12 |
+
fn_index=2
|
13 |
+
)
|
14 |
+
|
15 |
+
# Find the last occurrence of "."
|
16 |
+
last_period_index = fuyu_result.rfind('.')
|
17 |
+
|
18 |
+
# Truncate the string up to the last period
|
19 |
+
truncated_caption = fuyu_result[:last_period_index + 1]
|
20 |
+
|
21 |
+
# print(truncated_caption)
|
22 |
+
print(f"\n—\nIMAGE CAPTION: {truncated_caption}")
|
23 |
+
|
24 |
+
return truncated_caption
|
25 |
+
|
26 |
+
|
27 |
+
def infer(image_in):
|
28 |
+
gr.Info("Getting image caption with Fuyu...")
|
29 |
+
user_prompt = get_caption(image_in)
|
30 |
+
|
31 |
+
|
32 |
+
|
33 |
+
return user_prompt
|
34 |
+
|
35 |
+
title = f"LLM Agent from a Picture",
|
36 |
+
description = f"Get a LLM system prompt from a picture so you can use it in <a href='https://huggingface.co/spaces/abidlabs/GPT-Baker'>GPT-Baker</a>."
|
37 |
+
|
38 |
+
css = """
|
39 |
+
#col-container{
|
40 |
+
margin: 0 auto;
|
41 |
+
max-width: 780px;
|
42 |
+
text-align: left;
|
43 |
+
}
|
44 |
+
"""
|
45 |
+
|
46 |
+
with gr.Blocks(css=css) as demo:
|
47 |
+
with gr.Column(elem_id="col-container"):
|
48 |
+
gr.HTML(f"""
|
49 |
+
<h2 style="text-align: center;">LLM Agent from a Picture</h2>
|
50 |
+
<p style="text-align: center;">{description}</p>
|
51 |
+
""")
|
52 |
+
|
53 |
+
with gr.Row():
|
54 |
+
with gr.Column():
|
55 |
+
image_in = gr.Image(
|
56 |
+
label = "Image reference",
|
57 |
+
type = "filepath",
|
58 |
+
elem_id = "image-in"
|
59 |
+
)
|
60 |
+
submit_btn = gr.Button("Make desciptions of my pic !")
|
61 |
+
with gr.Column():
|
62 |
+
result = gr.Textbox(
|
63 |
+
label ="Suggested System",
|
64 |
+
lines = 6,
|
65 |
+
max_lines = 30,
|
66 |
+
elem_id = "suggested-system-prompt"
|
67 |
+
)
|
68 |
+
with gr.Row():
|
69 |
+
gr.Examples(
|
70 |
+
examples = [
|
71 |
+
["ponder.png"],
|
72 |
+
["ponder2.png"],
|
73 |
+
|
74 |
+
],
|
75 |
+
fn = infer,
|
76 |
+
inputs = [image_in],
|
77 |
+
outputs = [result],
|
78 |
+
cache_examples = True
|
79 |
+
)
|
80 |
+
|
81 |
+
submit_btn.click(
|
82 |
+
fn = infer,
|
83 |
+
inputs = [
|
84 |
+
image_in
|
85 |
+
],
|
86 |
+
outputs =[
|
87 |
+
result
|
88 |
+
]
|
89 |
+
)
|
90 |
+
|
91 |
+
demo.queue().launch(show_api=False)
|