Initial commit for themed Gradio app
Browse files- app.py +106 -0
- requirements.txt +3 -0
app.py
ADDED
@@ -0,0 +1,106 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
3 |
+
from PyPDF2 import PdfReader
|
4 |
+
|
5 |
+
# Models and tokenizers setup
|
6 |
+
models = {
|
7 |
+
"Text Generator (Bloom)": {
|
8 |
+
"model": AutoModelForSeq2SeqLM.from_pretrained("bigscience/bloom-560m"),
|
9 |
+
"tokenizer": AutoTokenizer.from_pretrained("bigscience/bloom-560m"),
|
10 |
+
},
|
11 |
+
"PDF Summarizer (T5)": {
|
12 |
+
"model": AutoModelForSeq2SeqLM.from_pretrained("t5-small"),
|
13 |
+
"tokenizer": AutoTokenizer.from_pretrained("t5-small"),
|
14 |
+
},
|
15 |
+
"Broken Answer (T0pp)": {
|
16 |
+
"model": AutoModelForSeq2SeqLM.from_pretrained("bigscience/T0pp"),
|
17 |
+
"tokenizer": AutoTokenizer.from_pretrained("bigscience/T0pp"),
|
18 |
+
},
|
19 |
+
}
|
20 |
+
|
21 |
+
# Function for text generation
|
22 |
+
def generate_text(model_choice, input_text, max_tokens, temperature, top_p):
|
23 |
+
model_info = models[model_choice]
|
24 |
+
tokenizer = model_info["tokenizer"]
|
25 |
+
model = model_info["model"]
|
26 |
+
|
27 |
+
inputs = tokenizer(input_text, return_tensors="pt", padding=True, truncation=True, max_length=512)
|
28 |
+
outputs = model.generate(
|
29 |
+
**inputs, max_length=max_tokens, num_beams=5, early_stopping=True, temperature=temperature, top_p=top_p
|
30 |
+
)
|
31 |
+
return tokenizer.decode(outputs[0], skip_special_tokens=True)
|
32 |
+
|
33 |
+
# Function for PDF summarization
|
34 |
+
def summarize_pdf(pdf_file, max_tokens, temperature, top_p):
|
35 |
+
reader = PdfReader(pdf_file)
|
36 |
+
text = ""
|
37 |
+
for page in reader.pages:
|
38 |
+
text += page.extract_text()
|
39 |
+
|
40 |
+
model_info = models["PDF Summarizer (T5)"]
|
41 |
+
tokenizer = model_info["tokenizer"]
|
42 |
+
model = model_info["model"]
|
43 |
+
|
44 |
+
inputs = tokenizer("summarize: " + text, return_tensors="pt", padding=True, truncation=True, max_length=512)
|
45 |
+
outputs = model.generate(
|
46 |
+
**inputs, max_length=max_tokens, num_beams=5, early_stopping=True, temperature=temperature, top_p=top_p
|
47 |
+
)
|
48 |
+
return tokenizer.decode(outputs[0], skip_special_tokens=True)
|
49 |
+
|
50 |
+
# Build Gradio interface
|
51 |
+
def launch_custom_app():
|
52 |
+
with gr.Blocks(theme=gr.themes.Monochrome()) as demo:
|
53 |
+
gr.Markdown("<h1 style='text-align: center;'>💡 Multi-Model Assistant</h1>")
|
54 |
+
gr.Markdown("<p style='text-align: center;'>Switch between text generation, PDF summarization, or quirky broken answers!</p>")
|
55 |
+
|
56 |
+
with gr.Tabs():
|
57 |
+
# Tab for Text Generation
|
58 |
+
with gr.Tab("Text Generator"):
|
59 |
+
model_choice = gr.Dropdown(choices=list(models.keys()), label="Choose a Model", value="Text Generator (Bloom)")
|
60 |
+
input_text = gr.Textbox(label="Enter Text")
|
61 |
+
max_tokens = gr.Slider(minimum=10, maximum=512, value=150, step=10, label="Max Tokens")
|
62 |
+
temperature = gr.Slider(minimum=0.1, maximum=1.0, value=0.7, step=0.1, label="Temperature")
|
63 |
+
top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)")
|
64 |
+
output_text = gr.Textbox(label="Generated Text", interactive=False)
|
65 |
+
generate_button = gr.Button("Generate Text")
|
66 |
+
|
67 |
+
generate_button.click(
|
68 |
+
generate_text,
|
69 |
+
inputs=[model_choice, input_text, max_tokens, temperature, top_p],
|
70 |
+
outputs=output_text
|
71 |
+
)
|
72 |
+
|
73 |
+
# Tab for PDF Summarization
|
74 |
+
with gr.Tab("PDF Summarizer"):
|
75 |
+
pdf_file = gr.File(label="Upload a PDF File", file_types=[".pdf"])
|
76 |
+
max_tokens_pdf = gr.Slider(minimum=10, maximum=512, value=150, step=10, label="Max Tokens")
|
77 |
+
temperature_pdf = gr.Slider(minimum=0.1, maximum=1.0, value=0.7, step=0.1, label="Temperature")
|
78 |
+
top_p_pdf = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)")
|
79 |
+
summary_output = gr.Textbox(label="PDF Summary", interactive=False)
|
80 |
+
summarize_button = gr.Button("Summarize PDF")
|
81 |
+
|
82 |
+
summarize_button.click(
|
83 |
+
summarize_pdf,
|
84 |
+
inputs=[pdf_file, max_tokens_pdf, temperature_pdf, top_p_pdf],
|
85 |
+
outputs=summary_output
|
86 |
+
)
|
87 |
+
|
88 |
+
# Tab for Broken Model
|
89 |
+
with gr.Tab("Broken Answers"):
|
90 |
+
broken_input = gr.Textbox(label="Enter Text")
|
91 |
+
broken_max_tokens = gr.Slider(minimum=10, maximum=512, value=150, step=10, label="Max Tokens")
|
92 |
+
broken_temperature = gr.Slider(minimum=0.1, maximum=1.0, value=0.7, step=0.1, label="Temperature")
|
93 |
+
broken_top_p = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)")
|
94 |
+
broken_output = gr.Textbox(label="Broken Model Output", interactive=False)
|
95 |
+
broken_button = gr.Button("Generate Broken Answer")
|
96 |
+
|
97 |
+
broken_button.click(
|
98 |
+
lambda text, max_tokens, temp, top_p: generate_text("Broken Answer (T0pp)", text, max_tokens, temp, top_p),
|
99 |
+
inputs=[broken_input, broken_max_tokens, broken_temperature, broken_top_p],
|
100 |
+
outputs=broken_output
|
101 |
+
)
|
102 |
+
|
103 |
+
demo.launch()
|
104 |
+
|
105 |
+
# Launch the app
|
106 |
+
launch_custom_app()
|
requirements.txt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
gradio
|
2 |
+
transformers
|
3 |
+
PyPDF2
|