Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -34,7 +34,7 @@ processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct", trust_rem
|
|
34 |
@spaces.GPU()
|
35 |
def process_pdf_and_query(pdf_file, user_query):
|
36 |
# Convert the PDF to images
|
37 |
-
images = convert_from_path(pdf_file.name)
|
38 |
num_images = len(images)
|
39 |
|
40 |
# Indexing the PDF in RAG
|
@@ -88,12 +88,26 @@ def process_pdf_and_query(pdf_file, user_query):
|
|
88 |
|
89 |
return output_text[0], num_images
|
90 |
|
|
|
91 |
pdf_input = gr.File(label="Upload PDF")
|
92 |
query_input = gr.Textbox(label="Enter your query", placeholder="Ask a question about the PDF")
|
93 |
-
output_text = gr.Textbox(label="Model Answer")
|
94 |
output_images = gr.Textbox(label="Number of Images in PDF")
|
95 |
|
96 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
97 |
footer = """
|
98 |
<div style="text-align: center; margin-top: 20px;">
|
99 |
<a href="https://www.linkedin.com/in/pejman-ebrahimi-4a60151a7/" target="_blank">LinkedIn</a> |
|
@@ -107,18 +121,6 @@ footer = """
|
|
107 |
</div>
|
108 |
"""
|
109 |
|
110 |
-
explanation = """
|
111 |
-
<div style="text-align: center; margin-bottom: 20px;">
|
112 |
-
<h2 style="font-weight: bold; font-size: 24px;">Multimodal RAG (Retrieval-Augmented Generation)</h2>
|
113 |
-
<p>
|
114 |
-
This application utilizes the ColPali model as a multimodal retriever,
|
115 |
-
which retrieves relevant information from documents and generates answers
|
116 |
-
using the Qwen/Qwen2-VL-2B-Instruct LLM (Large Language Model)
|
117 |
-
via the Byaldi library, developed by Answer.ai.
|
118 |
-
</p>
|
119 |
-
</div>
|
120 |
-
"""
|
121 |
-
|
122 |
|
123 |
demo = gr.Interface(
|
124 |
fn=process_pdf_and_query,
|
@@ -130,10 +132,12 @@ demo = gr.Interface(
|
|
130 |
|
131 |
|
132 |
with demo:
|
133 |
-
gr.HTML(explanation)
|
134 |
-
gr.HTML(footer)
|
135 |
gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
|
136 |
-
|
|
|
|
|
137 |
|
138 |
css = """
|
139 |
<style>
|
@@ -145,6 +149,7 @@ css = """
|
|
145 |
padding: 10px 20px;
|
146 |
font-size: 16px;
|
147 |
cursor: pointer;
|
|
|
148 |
}
|
149 |
.duplicate-button {
|
150 |
background-color: green;
|
@@ -160,4 +165,5 @@ css = """
|
|
160 |
"""
|
161 |
gr.HTML(css)
|
162 |
|
163 |
-
|
|
|
|
34 |
@spaces.GPU()
|
35 |
def process_pdf_and_query(pdf_file, user_query):
|
36 |
# Convert the PDF to images
|
37 |
+
images = convert_from_path(pdf_file.name)
|
38 |
num_images = len(images)
|
39 |
|
40 |
# Indexing the PDF in RAG
|
|
|
88 |
|
89 |
return output_text[0], num_images
|
90 |
|
91 |
+
|
92 |
pdf_input = gr.File(label="Upload PDF")
|
93 |
query_input = gr.Textbox(label="Enter your query", placeholder="Ask a question about the PDF")
|
94 |
+
output_text = gr.Textbox(label="Model Answer")
|
95 |
output_images = gr.Textbox(label="Number of Images in PDF")
|
96 |
|
97 |
|
98 |
+
explanation = """
|
99 |
+
<div style="text-align: center; margin-bottom: 20px;">
|
100 |
+
<h2 style="font-weight: bold; font-size: 24px;">Multimodal RAG (Retrieval-Augmented Generation)</h2>
|
101 |
+
<p>
|
102 |
+
This application utilizes the ColPali model as a multimodal retriever,
|
103 |
+
which retrieves relevant information from documents and generates answers
|
104 |
+
using the Qwen/Qwen2-VL-2B-Instruct LLM (Large Language Model)
|
105 |
+
via the Byaldi library, developed by Answer.ai.
|
106 |
+
</p>
|
107 |
+
</div>
|
108 |
+
"""
|
109 |
+
|
110 |
+
|
111 |
footer = """
|
112 |
<div style="text-align: center; margin-top: 20px;">
|
113 |
<a href="https://www.linkedin.com/in/pejman-ebrahimi-4a60151a7/" target="_blank">LinkedIn</a> |
|
|
|
121 |
</div>
|
122 |
"""
|
123 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
124 |
|
125 |
demo = gr.Interface(
|
126 |
fn=process_pdf_and_query,
|
|
|
132 |
|
133 |
|
134 |
with demo:
|
135 |
+
gr.HTML(explanation)
|
136 |
+
gr.HTML(footer)
|
137 |
gr.DuplicateButton(value="Duplicate Space for private use", elem_classes="duplicate-button")
|
138 |
+
|
139 |
+
submit_btn = gr.Button("Submit", elem_classes="submit-button")
|
140 |
+
|
141 |
|
142 |
css = """
|
143 |
<style>
|
|
|
149 |
padding: 10px 20px;
|
150 |
font-size: 16px;
|
151 |
cursor: pointer;
|
152 |
+
margin: 10px; /* Add some space between buttons */
|
153 |
}
|
154 |
.duplicate-button {
|
155 |
background-color: green;
|
|
|
165 |
"""
|
166 |
gr.HTML(css)
|
167 |
|
168 |
+
# Launch the Gradio app
|
169 |
+
demo.launch(debug=True)
|