Spaces:
Running
Running
app.py
CHANGED
@@ -1,13 +1,16 @@
|
|
1 |
import gradio as gr
|
2 |
import re
|
|
|
3 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
4 |
from nltk.corpus import stopwords
|
5 |
-
|
6 |
-
# Download the NLTK stopwords (only the first time you run)
|
7 |
import nltk
|
|
|
|
|
8 |
nltk.download('stopwords')
|
|
|
9 |
|
10 |
-
# Model choices
|
11 |
model_choices = {
|
12 |
"Pegasus (google/pegasus-xsum)": "google/pegasus-xsum",
|
13 |
"BigBird-Pegasus (google/bigbird-pegasus-large-arxiv)": "google/bigbird-pegasus-large-arxiv",
|
@@ -33,66 +36,54 @@ model_choices = {
|
|
33 |
|
34 |
model_cache = {}
|
35 |
|
36 |
-
# Get NLTK stopwords (common stop words)
|
37 |
-
stop_words = set(stopwords.words('english'))
|
38 |
-
|
39 |
-
# Function to clean input text by removing unnecessary words like stop words
|
40 |
def clean_text(input_text):
|
41 |
-
# Replace special characters with a space
|
42 |
cleaned_text = re.sub(r'[^A-Za-z0-9\s]', ' ', input_text)
|
43 |
-
|
44 |
-
# Tokenize the input text and remove stop words
|
45 |
words = cleaned_text.split()
|
46 |
words = [word for word in words if word.lower() not in stop_words]
|
47 |
-
|
48 |
-
# Rebuild the cleaned text
|
49 |
-
cleaned_text = " ".join(words)
|
50 |
-
|
51 |
-
# Strip leading and trailing spaces
|
52 |
-
cleaned_text = cleaned_text.strip()
|
53 |
-
|
54 |
-
return cleaned_text
|
55 |
|
56 |
-
# Load model and tokenizer
|
57 |
def load_model(model_name):
|
58 |
if model_name not in model_cache:
|
59 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
60 |
-
model = AutoModelForSeq2SeqLM.from_pretrained(
|
|
|
|
|
|
|
|
|
|
|
61 |
model_cache[model_name] = (tokenizer, model)
|
|
|
|
|
|
|
|
|
|
|
62 |
return model_cache[model_name]
|
63 |
|
64 |
-
#
|
65 |
def summarize_text(input_text, model_label, char_limit):
|
66 |
if not input_text.strip():
|
67 |
return "Please enter some text."
|
68 |
|
69 |
-
# Clean the input text by removing special characters and stop words
|
70 |
input_text = clean_text(input_text)
|
71 |
-
|
72 |
model_name = model_choices[model_label]
|
73 |
tokenizer, model = load_model(model_name)
|
74 |
|
75 |
-
# Adjust the input format for T5 and FLAN models
|
76 |
if "t5" in model_name.lower() or "flan" in model_name.lower():
|
77 |
input_text = "summarize: " + input_text
|
78 |
|
|
|
79 |
inputs = tokenizer(input_text, return_tensors="pt", truncation=True)
|
|
|
80 |
|
81 |
summary_ids = model.generate(
|
82 |
-
|
83 |
-
max_length=
|
84 |
min_length=5,
|
85 |
do_sample=False
|
86 |
)
|
87 |
|
88 |
-
# Decode the summary while skipping special tokens and cleaning unwanted characters
|
89 |
summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
|
90 |
-
|
91 |
-
# Remove unwanted characters like pipes or any unwanted symbols
|
92 |
-
summary = summary.replace("|", "") # Remove pipes
|
93 |
-
summary = summary.strip() # Remove leading/trailing whitespace
|
94 |
-
|
95 |
-
return summary[:char_limit] # Enforce character limit
|
96 |
|
97 |
# Gradio UI
|
98 |
iface = gr.Interface(
|
@@ -100,11 +91,11 @@ iface = gr.Interface(
|
|
100 |
inputs=[
|
101 |
gr.Textbox(lines=6, label="Enter text to summarize"),
|
102 |
gr.Dropdown(choices=list(model_choices.keys()), label="Choose summarization model", value="Pegasus (google/pegasus-xsum)"),
|
103 |
-
gr.Slider(minimum=30, maximum=200, value=
|
104 |
],
|
105 |
outputs=gr.Textbox(lines=3, label="Summary (truncated to character limit)"),
|
106 |
-
title="Multi-Model Text Summarizer",
|
107 |
-
description="Summarize
|
108 |
)
|
109 |
|
110 |
iface.launch()
|
|
|
1 |
import gradio as gr
|
2 |
import re
|
3 |
+
import torch
|
4 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
5 |
from nltk.corpus import stopwords
|
6 |
+
from spaces import GPU
|
|
|
7 |
import nltk
|
8 |
+
|
9 |
+
# Download stopwords if not already
|
10 |
nltk.download('stopwords')
|
11 |
+
stop_words = set(stopwords.words('english'))
|
12 |
|
13 |
+
# Model choices
|
14 |
model_choices = {
|
15 |
"Pegasus (google/pegasus-xsum)": "google/pegasus-xsum",
|
16 |
"BigBird-Pegasus (google/bigbird-pegasus-large-arxiv)": "google/bigbird-pegasus-large-arxiv",
|
|
|
36 |
|
37 |
model_cache = {}
|
38 |
|
|
|
|
|
|
|
|
|
39 |
def clean_text(input_text):
|
|
|
40 |
cleaned_text = re.sub(r'[^A-Za-z0-9\s]', ' ', input_text)
|
|
|
|
|
41 |
words = cleaned_text.split()
|
42 |
words = [word for word in words if word.lower() not in stop_words]
|
43 |
+
return " ".join(words).strip()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
44 |
|
|
|
45 |
def load_model(model_name):
|
46 |
if model_name not in model_cache:
|
47 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
48 |
+
model = AutoModelForSeq2SeqLM.from_pretrained(
|
49 |
+
model_name,
|
50 |
+
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32
|
51 |
+
)
|
52 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
53 |
+
model = model.to(device)
|
54 |
model_cache[model_name] = (tokenizer, model)
|
55 |
+
|
56 |
+
# Warm up with dummy input
|
57 |
+
dummy_input = tokenizer("summarize: hello world", return_tensors="pt").input_ids.to(device)
|
58 |
+
model.generate(dummy_input, max_length=10)
|
59 |
+
|
60 |
return model_cache[model_name]
|
61 |
|
62 |
+
@GPU # 👈 Required for ZeroGPU to allocate GPU when this is called
|
63 |
def summarize_text(input_text, model_label, char_limit):
|
64 |
if not input_text.strip():
|
65 |
return "Please enter some text."
|
66 |
|
|
|
67 |
input_text = clean_text(input_text)
|
|
|
68 |
model_name = model_choices[model_label]
|
69 |
tokenizer, model = load_model(model_name)
|
70 |
|
|
|
71 |
if "t5" in model_name.lower() or "flan" in model_name.lower():
|
72 |
input_text = "summarize: " + input_text
|
73 |
|
74 |
+
device = model.device
|
75 |
inputs = tokenizer(input_text, return_tensors="pt", truncation=True)
|
76 |
+
input_ids = inputs["input_ids"].to(device)
|
77 |
|
78 |
summary_ids = model.generate(
|
79 |
+
input_ids,
|
80 |
+
max_length=30,
|
81 |
min_length=5,
|
82 |
do_sample=False
|
83 |
)
|
84 |
|
|
|
85 |
summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
|
86 |
+
return summary[:char_limit].strip()
|
|
|
|
|
|
|
|
|
|
|
87 |
|
88 |
# Gradio UI
|
89 |
iface = gr.Interface(
|
|
|
91 |
inputs=[
|
92 |
gr.Textbox(lines=6, label="Enter text to summarize"),
|
93 |
gr.Dropdown(choices=list(model_choices.keys()), label="Choose summarization model", value="Pegasus (google/pegasus-xsum)"),
|
94 |
+
gr.Slider(minimum=30, maximum=200, value=80, step=1, label="Max Character Limit")
|
95 |
],
|
96 |
outputs=gr.Textbox(lines=3, label="Summary (truncated to character limit)"),
|
97 |
+
title="Multi-Model Text Summarizer (GPU Ready)",
|
98 |
+
description="Summarize long or short texts using state-of-the-art Hugging Face models with GPU acceleration (ZeroGPU-compatible)."
|
99 |
)
|
100 |
|
101 |
iface.launch()
|