Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,28 +1,29 @@
|
|
1 |
import gradio as gr
|
2 |
-
from groq import Groq
|
3 |
import os
|
4 |
import threading
|
5 |
import base64
|
6 |
from io import BytesIO
|
|
|
7 |
|
8 |
-
# Initialize Groq
|
9 |
-
|
10 |
|
11 |
-
# Load Text-to-Image Models
|
12 |
model1 = gr.load("models/prithivMLmods/SD3.5-Turbo-Realism-2.0-LoRA")
|
13 |
model2 = gr.load("models/Purz/face-projection")
|
|
|
14 |
|
15 |
-
# Stop
|
16 |
stop_event = threading.Event()
|
17 |
|
18 |
-
# Convert PIL image to Base64
|
19 |
-
def pil_to_base64(pil_image, image_format=
|
20 |
buffered = BytesIO()
|
21 |
pil_image.save(buffered, format=image_format)
|
22 |
-
base64_string = base64.b64encode(buffered.getvalue()).decode(
|
23 |
return base64_string, image_format
|
24 |
|
25 |
-
# Function for Visual Question Answering (
|
26 |
def answer_question(text, image, temperature=0.0, max_tokens=1024):
|
27 |
base64_string, file_format = pil_to_base64(image)
|
28 |
|
@@ -36,8 +37,8 @@ def answer_question(text, image, temperature=0.0, max_tokens=1024):
|
|
36 |
}
|
37 |
]
|
38 |
|
39 |
-
chat_response =
|
40 |
-
model="
|
41 |
messages=messages,
|
42 |
temperature=temperature,
|
43 |
max_tokens=max_tokens
|
@@ -45,18 +46,24 @@ def answer_question(text, image, temperature=0.0, max_tokens=1024):
|
|
45 |
|
46 |
return chat_response.choices[0].message.content
|
47 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
|
49 |
-
# Clear
|
50 |
def clear_all():
|
51 |
-
return "", None, ""
|
52 |
-
|
53 |
|
54 |
-
# Set up
|
55 |
with gr.Blocks() as demo:
|
56 |
-
gr.Markdown("# ๐ AI Tutor &
|
57 |
|
58 |
-
# Section
|
59 |
-
gr.Markdown("## ๐ผ๏ธ Visual Question Answering (
|
60 |
with gr.Row():
|
61 |
with gr.Column(scale=2):
|
62 |
question = gr.Textbox(placeholder="Ask about the image...", lines=2)
|
@@ -66,24 +73,42 @@ with gr.Blocks() as demo:
|
|
66 |
max_tokens = gr.Slider(label="Max Tokens", minimum=128, maximum=2048, value=1024, step=128)
|
67 |
|
68 |
with gr.Column(scale=3):
|
69 |
-
output_text = gr.Textbox(lines=10, label="
|
70 |
|
71 |
with gr.Row():
|
72 |
clear_btn = gr.Button("Clear", variant="secondary")
|
73 |
submit_btn_vqa = gr.Button("Submit", variant="primary")
|
74 |
|
75 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
76 |
submit_btn_vqa.click(
|
77 |
fn=answer_question,
|
78 |
inputs=[question, image, temperature, max_tokens],
|
79 |
outputs=[output_text]
|
80 |
)
|
81 |
|
82 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
83 |
clear_btn.click(
|
84 |
fn=clear_all,
|
85 |
inputs=[],
|
86 |
-
outputs=[question, image, output_text]
|
87 |
)
|
88 |
|
89 |
if __name__ == "__main__":
|
|
|
1 |
import gradio as gr
|
|
|
2 |
import os
|
3 |
import threading
|
4 |
import base64
|
5 |
from io import BytesIO
|
6 |
+
from groq import Groq
|
7 |
|
8 |
+
# ๐น Initialize Groq API Client (FREE)
|
9 |
+
groq_client = Groq(api_key=os.getenv("GROQ_API_KEY"))
|
10 |
|
11 |
+
# ๐น Load Text-to-Image Models (Restoring Multi-Image Generation)
|
12 |
model1 = gr.load("models/prithivMLmods/SD3.5-Turbo-Realism-2.0-LoRA")
|
13 |
model2 = gr.load("models/Purz/face-projection")
|
14 |
+
model3 = gr.load("models/stablediffusion/stable-diffusion-xl")
|
15 |
|
16 |
+
# ๐น Stop Event for Threading
|
17 |
stop_event = threading.Event()
|
18 |
|
19 |
+
# ๐น Convert PIL image to Base64
|
20 |
+
def pil_to_base64(pil_image, image_format="jpeg"):
|
21 |
buffered = BytesIO()
|
22 |
pil_image.save(buffered, format=image_format)
|
23 |
+
base64_string = base64.b64encode(buffered.getvalue()).decode("utf-8")
|
24 |
return base64_string, image_format
|
25 |
|
26 |
+
# ๐น Function for Visual Question Answering (VQA) with Mixtral-8x7B
|
27 |
def answer_question(text, image, temperature=0.0, max_tokens=1024):
|
28 |
base64_string, file_format = pil_to_base64(image)
|
29 |
|
|
|
37 |
}
|
38 |
]
|
39 |
|
40 |
+
chat_response = groq_client.chat.completions.create(
|
41 |
+
model="mixtral-8x7b-32768",
|
42 |
messages=messages,
|
43 |
temperature=temperature,
|
44 |
max_tokens=max_tokens
|
|
|
46 |
|
47 |
return chat_response.choices[0].message.content
|
48 |
|
49 |
+
# ๐น Function to Generate Three Images (Multi-Output)
|
50 |
+
def generate_images(prompt):
|
51 |
+
stop_event.clear()
|
52 |
+
img1 = model1.predict(prompt)
|
53 |
+
img2 = model2.predict(prompt)
|
54 |
+
img3 = model3.predict(prompt)
|
55 |
+
return img1, img2, img3
|
56 |
|
57 |
+
# ๐น Clear All Fields
|
58 |
def clear_all():
|
59 |
+
return "", None, "", None, None, None
|
|
|
60 |
|
61 |
+
# ๐น Set up Gradio Interface
|
62 |
with gr.Blocks() as demo:
|
63 |
+
gr.Markdown("# ๐ AI Tutor, VQA & Image Generation")
|
64 |
|
65 |
+
# ๐น Section 1: Visual Question Answering (Groq)
|
66 |
+
gr.Markdown("## ๐ผ๏ธ Visual Question Answering (Mixtral-8x7B)")
|
67 |
with gr.Row():
|
68 |
with gr.Column(scale=2):
|
69 |
question = gr.Textbox(placeholder="Ask about the image...", lines=2)
|
|
|
73 |
max_tokens = gr.Slider(label="Max Tokens", minimum=128, maximum=2048, value=1024, step=128)
|
74 |
|
75 |
with gr.Column(scale=3):
|
76 |
+
output_text = gr.Textbox(lines=10, label="Mixtral VQA Response")
|
77 |
|
78 |
with gr.Row():
|
79 |
clear_btn = gr.Button("Clear", variant="secondary")
|
80 |
submit_btn_vqa = gr.Button("Submit", variant="primary")
|
81 |
|
82 |
+
# ๐น Section 2: Image Generation (3 Outputs)
|
83 |
+
gr.Markdown("## ๐จ AI-Generated Images (3 Variations)")
|
84 |
+
with gr.Row():
|
85 |
+
prompt = gr.Textbox(placeholder="Describe the image you want...", lines=2)
|
86 |
+
generate_btn = gr.Button("Generate Images", variant="primary")
|
87 |
+
|
88 |
+
with gr.Row():
|
89 |
+
image1 = gr.Image(label="Image 1")
|
90 |
+
image2 = gr.Image(label="Image 2")
|
91 |
+
image3 = gr.Image(label="Image 3")
|
92 |
+
|
93 |
+
# ๐น VQA Processing
|
94 |
submit_btn_vqa.click(
|
95 |
fn=answer_question,
|
96 |
inputs=[question, image, temperature, max_tokens],
|
97 |
outputs=[output_text]
|
98 |
)
|
99 |
|
100 |
+
# ๐น Image Generation Processing
|
101 |
+
generate_btn.click(
|
102 |
+
fn=generate_images,
|
103 |
+
inputs=[prompt],
|
104 |
+
outputs=[image1, image2, image3]
|
105 |
+
)
|
106 |
+
|
107 |
+
# ๐น Clear All Inputs
|
108 |
clear_btn.click(
|
109 |
fn=clear_all,
|
110 |
inputs=[],
|
111 |
+
outputs=[question, image, output_text, image1, image2, image3]
|
112 |
)
|
113 |
|
114 |
if __name__ == "__main__":
|