Spaces:
Runtime error
Runtime error
modularized code
Browse files- app.py +28 -34
- src/abstractive_summarizer.py +22 -0
- src/vanilla_summarizer.py +0 -0
app.py
CHANGED
|
@@ -1,35 +1,19 @@
|
|
| 1 |
import torch
|
| 2 |
import streamlit as st
|
| 3 |
-
from
|
| 4 |
-
from transformers import T5Tokenizer, T5ForConditionalGeneration, T5Config
|
| 5 |
-
|
| 6 |
-
def abstractive_summarizer(text : str, model):
|
| 7 |
-
tokenizer = T5Tokenizer.from_pretrained('t5-large')
|
| 8 |
-
device = torch.device('cpu')
|
| 9 |
-
preprocess_text = text.strip().replace("\n", "")
|
| 10 |
-
t5_prepared_text = "summarize: " + preprocess_text
|
| 11 |
-
tokenized_text = tokenizer.encode(t5_prepared_text, return_tensors="pt").to(device)
|
| 12 |
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
min_length=30,
|
| 18 |
-
max_length=100,
|
| 19 |
-
early_stopping=True)
|
| 20 |
-
abs_summarized_text = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
|
| 21 |
-
|
| 22 |
-
return abs_summarized_text
|
| 23 |
|
| 24 |
-
# @st.cache()
|
| 25 |
-
# def load_ext_model():
|
| 26 |
-
# model = Summarizer()
|
| 27 |
-
# return model
|
| 28 |
|
|
|
|
| 29 |
@st.cache()
|
| 30 |
def load_abs_model():
|
| 31 |
-
|
| 32 |
-
|
|
|
|
| 33 |
|
| 34 |
|
| 35 |
if __name__ == "__main__":
|
|
@@ -37,10 +21,14 @@ if __name__ == "__main__":
|
|
| 37 |
# Main Application
|
| 38 |
# ---------------------------------
|
| 39 |
st.title("Text Summarizer π")
|
| 40 |
-
summarize_type = st.sidebar.selectbox(
|
|
|
|
|
|
|
| 41 |
|
| 42 |
inp_text = st.text_input("Enter the text here")
|
| 43 |
|
|
|
|
|
|
|
| 44 |
# view summarized text (expander)
|
| 45 |
with st.expander("View input text"):
|
| 46 |
st.write(inp_text)
|
|
@@ -51,16 +39,22 @@ if __name__ == "__main__":
|
|
| 51 |
if summarize:
|
| 52 |
if summarize_type == "Extractive":
|
| 53 |
# extractive summarizer
|
| 54 |
-
|
| 55 |
-
with st.spinner(
|
|
|
|
|
|
|
| 56 |
ext_model = Summarizer()
|
| 57 |
summarized_text = ext_model(inp_text, num_sentences=5)
|
| 58 |
-
|
| 59 |
-
elif summarize_type == "Abstractive":
|
| 60 |
-
with st.spinner(text="Creating abstractive summary. This might take a few seconds ..."):
|
| 61 |
-
abs_model = load_abs_model()
|
| 62 |
-
summarized_text = abstractive_summarizer(inp_text, model=abs_model)
|
| 63 |
|
| 64 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 65 |
st.subheader("Summarized text")
|
| 66 |
st.info(summarized_text)
|
|
|
|
| 1 |
import torch
|
| 2 |
import streamlit as st
|
| 3 |
+
from transformers import T5Tokenizer, T5ForConditionalGeneration
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
|
| 5 |
+
# local modules
|
| 6 |
+
from extractive_summarizer.model_processors import Summarizer
|
| 7 |
+
from src.utils import clean_text
|
| 8 |
+
from src.abstractive_summarizer import abstractive_summarizer
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
+
# abstractive summarizer model
|
| 12 |
@st.cache()
|
| 13 |
def load_abs_model():
|
| 14 |
+
tokenizer = T5Tokenizer.from_pretrained("t5-large")
|
| 15 |
+
model = T5ForConditionalGeneration.from_pretrained("t5-base")
|
| 16 |
+
return tokenizer, model
|
| 17 |
|
| 18 |
|
| 19 |
if __name__ == "__main__":
|
|
|
|
| 21 |
# Main Application
|
| 22 |
# ---------------------------------
|
| 23 |
st.title("Text Summarizer π")
|
| 24 |
+
summarize_type = st.sidebar.selectbox(
|
| 25 |
+
"Summarization type", options=["Extractive", "Abstractive"]
|
| 26 |
+
)
|
| 27 |
|
| 28 |
inp_text = st.text_input("Enter the text here")
|
| 29 |
|
| 30 |
+
inp_text = clean_text(inp_text)
|
| 31 |
+
|
| 32 |
# view summarized text (expander)
|
| 33 |
with st.expander("View input text"):
|
| 34 |
st.write(inp_text)
|
|
|
|
| 39 |
if summarize:
|
| 40 |
if summarize_type == "Extractive":
|
| 41 |
# extractive summarizer
|
| 42 |
+
|
| 43 |
+
with st.spinner(
|
| 44 |
+
text="Creating extractive summary. This might take a few seconds ..."
|
| 45 |
+
):
|
| 46 |
ext_model = Summarizer()
|
| 47 |
summarized_text = ext_model(inp_text, num_sentences=5)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
|
| 49 |
+
elif summarize_type == "Abstractive":
|
| 50 |
+
with st.spinner(
|
| 51 |
+
text="Creating abstractive summary. This might take a few seconds ..."
|
| 52 |
+
):
|
| 53 |
+
abs_tokenizer, abs_model = load_abs_model()
|
| 54 |
+
summarized_text = abstractive_summarizer(
|
| 55 |
+
abs_tokenizer, abs_model, inp_text
|
| 56 |
+
)
|
| 57 |
+
|
| 58 |
+
# final summarized output
|
| 59 |
st.subheader("Summarized text")
|
| 60 |
st.info(summarized_text)
|
src/abstractive_summarizer.py
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import torch
|
| 2 |
+
from transformers import T5Tokenizer
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
def abstractive_summarizer(tokenizer, model, text):
|
| 6 |
+
device = torch.device("cpu")
|
| 7 |
+
preprocess_text = text.strip().replace("\n", "")
|
| 8 |
+
t5_prepared_text = "summarize: " + preprocess_text
|
| 9 |
+
tokenized_text = tokenizer.encode(t5_prepared_text, return_tensors="pt").to(device)
|
| 10 |
+
|
| 11 |
+
# summmarize
|
| 12 |
+
summary_ids = model.generate(
|
| 13 |
+
tokenized_text,
|
| 14 |
+
num_beams=4,
|
| 15 |
+
no_repeat_ngram_size=2,
|
| 16 |
+
min_length=30,
|
| 17 |
+
max_length=100,
|
| 18 |
+
early_stopping=True,
|
| 19 |
+
)
|
| 20 |
+
abs_summarized_text = tokenizer.decode(summary_ids[0], skip_special_tokens=True)
|
| 21 |
+
|
| 22 |
+
return abs_summarized_text
|
src/vanilla_summarizer.py
DELETED
|
File without changes
|