Spaces:
Runtime error
Runtime error
Commit ·
960672d
1
Parent(s): 5b84a58
interface created
Browse files- README.md +3 -1
- app.py +100 -0
- requirements.txt +96 -0
- skimlit_max_trained_model/keras_metadata.pb +3 -0
- skimlit_max_trained_model/saved_model.pb +3 -0
- skimlit_max_trained_model/variables/variables.index +0 -0
README.md
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
---
|
| 2 |
title: Skimlit
|
| 3 |
-
emoji:
|
| 4 |
colorFrom: red
|
| 5 |
colorTo: red
|
| 6 |
sdk: gradio
|
|
@@ -10,3 +10,5 @@ pinned: false
|
|
| 10 |
---
|
| 11 |
|
| 12 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
title: Skimlit
|
| 3 |
+
emoji: 📄🔥
|
| 4 |
colorFrom: red
|
| 5 |
colorTo: red
|
| 6 |
sdk: gradio
|
|
|
|
| 10 |
---
|
| 11 |
|
| 12 |
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
| 13 |
+
|
| 14 |
+
Original source code can be found at https://www.github.com/harisj58/skimlit
|
app.py
ADDED
|
@@ -0,0 +1,100 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import tensorflow as tf
|
| 3 |
+
from keras.layers.preprocessing.text_vectorization import TextVectorization
|
| 4 |
+
import pandas as pd
|
| 5 |
+
import re, string
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
model_path = "skimlit_max_trained_model"
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
@tf.keras.utils.register_keras_serializable(package="Custom", name=None)
|
| 12 |
+
def custom_standardization(input_data):
|
| 13 |
+
lowercase = tf.strings.lower(input_data)
|
| 14 |
+
stripped_html = tf.strings.regex_replace(lowercase, "", " ")
|
| 15 |
+
return tf.strings.regex_replace(
|
| 16 |
+
stripped_html, "[%s]" % re.escape(string.punctuation), ""
|
| 17 |
+
)
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
model = None
|
| 21 |
+
with tf.keras.utils.custom_object_scope(
|
| 22 |
+
{"custom_standardization": custom_standardization}
|
| 23 |
+
):
|
| 24 |
+
model = tf.keras.models.load_model(model_path) # Folder
|
| 25 |
+
|
| 26 |
+
if model is not None:
|
| 27 |
+
model.summary()
|
| 28 |
+
else:
|
| 29 |
+
print("Model not loaded, sorry...")
|
| 30 |
+
|
| 31 |
+
# Load saved model
|
| 32 |
+
loaded_model = tf.keras.models.load_model(model_path)
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
def get_results(abs_text):
|
| 36 |
+
text = ""
|
| 37 |
+
labels = ["BACKGROUND", "CONCLUSIONS", "METHODS", "OBJECTIVE", "RESULTS"]
|
| 38 |
+
sent_list = abs_text.split(sep=".")
|
| 39 |
+
sent_list.pop()
|
| 40 |
+
sent_list = [sentence.strip() for sentence in sent_list]
|
| 41 |
+
i = 0
|
| 42 |
+
total_lines = len(sent_list)
|
| 43 |
+
final_list = []
|
| 44 |
+
temp = {}
|
| 45 |
+
for line in sent_list:
|
| 46 |
+
temp["text"] = line
|
| 47 |
+
temp["line_number"] = i
|
| 48 |
+
temp["total_lines"] = total_lines
|
| 49 |
+
i += 1
|
| 50 |
+
final_list.append(temp)
|
| 51 |
+
temp = {}
|
| 52 |
+
df = pd.DataFrame(final_list)
|
| 53 |
+
chars = [" ".join(list(sentence)) for sentence in sent_list]
|
| 54 |
+
line_numbers_one_hot = tf.one_hot(df.line_number.to_numpy(), depth=15)
|
| 55 |
+
lines_total_one_hot = tf.one_hot(df.total_lines.to_numpy(), depth=20)
|
| 56 |
+
preds = tf.argmax(
|
| 57 |
+
loaded_model.predict(
|
| 58 |
+
x=(
|
| 59 |
+
line_numbers_one_hot,
|
| 60 |
+
lines_total_one_hot,
|
| 61 |
+
tf.constant(sent_list),
|
| 62 |
+
tf.constant(chars),
|
| 63 |
+
),
|
| 64 |
+
verbose=0,
|
| 65 |
+
),
|
| 66 |
+
axis=1,
|
| 67 |
+
)
|
| 68 |
+
i = 0
|
| 69 |
+
for i in range(0, total_lines):
|
| 70 |
+
if i != 0 and preds[i] != preds[i - 1]:
|
| 71 |
+
text += f"\n\n{labels[preds[i]]}: "
|
| 72 |
+
text += f"{sent_list[i]}."
|
| 73 |
+
# print(f"\n\n{labels[preds[i]]}:", end=" ")
|
| 74 |
+
# print(f"{sent_list[i]}.", end=" ")
|
| 75 |
+
elif i == 0:
|
| 76 |
+
text += f"{labels[preds[i]]}: "
|
| 77 |
+
text += f"{sent_list[i]}."
|
| 78 |
+
# print(f"{labels[preds[i]]}:", end=" ")
|
| 79 |
+
# print(f"{sent_list[i]}.", end=" ")
|
| 80 |
+
else:
|
| 81 |
+
text += f"{sent_list[i]}."
|
| 82 |
+
# print(f"{sent_list[i]}.", end=" ")
|
| 83 |
+
return text
|
| 84 |
+
|
| 85 |
+
|
| 86 |
+
def skimlit_analysis(text):
|
| 87 |
+
# print(output)
|
| 88 |
+
return get_results(text)
|
| 89 |
+
|
| 90 |
+
|
| 91 |
+
intfc = gr.Interface(
|
| 92 |
+
fn=skimlit_analysis,
|
| 93 |
+
inputs=gr.Textbox(
|
| 94 |
+
label="Input abstract here",
|
| 95 |
+
lines=2,
|
| 96 |
+
placeholder="Input your abstract here (the model works best with ones from medical research papers)",
|
| 97 |
+
),
|
| 98 |
+
outputs=gr.Text(label="Skimmable Abstract", interactive=False),
|
| 99 |
+
)
|
| 100 |
+
intfc.launch()
|
requirements.txt
ADDED
|
@@ -0,0 +1,96 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
absl-py==1.4.0
|
| 2 |
+
aiofiles==23.1.0
|
| 3 |
+
aiohttp==3.8.4
|
| 4 |
+
aiosignal==1.3.1
|
| 5 |
+
altair==5.0.0
|
| 6 |
+
anyio==3.6.2
|
| 7 |
+
astunparse==1.6.3
|
| 8 |
+
async-timeout==4.0.2
|
| 9 |
+
attrs==23.1.0
|
| 10 |
+
cachetools==5.3.0
|
| 11 |
+
certifi==2023.5.7
|
| 12 |
+
charset-normalizer==3.1.0
|
| 13 |
+
click==8.1.3
|
| 14 |
+
colorama==0.4.6
|
| 15 |
+
contourpy==1.0.7
|
| 16 |
+
cycler==0.11.0
|
| 17 |
+
fastapi==0.95.1
|
| 18 |
+
ffmpy==0.3.0
|
| 19 |
+
filelock==3.12.0
|
| 20 |
+
flatbuffers==23.5.9
|
| 21 |
+
fonttools==4.39.4
|
| 22 |
+
frozenlist==1.3.3
|
| 23 |
+
fsspec==2023.5.0
|
| 24 |
+
gast==0.4.0
|
| 25 |
+
google-auth==2.18.0
|
| 26 |
+
google-auth-oauthlib==1.0.0
|
| 27 |
+
google-pasta==0.2.0
|
| 28 |
+
gradio==3.30.0
|
| 29 |
+
gradio_client==0.2.4
|
| 30 |
+
grpcio==1.54.2
|
| 31 |
+
h11==0.14.0
|
| 32 |
+
h5py==3.8.0
|
| 33 |
+
httpcore==0.17.0
|
| 34 |
+
httpx==0.24.0
|
| 35 |
+
huggingface-hub==0.14.1
|
| 36 |
+
idna==3.4
|
| 37 |
+
jax==0.4.10
|
| 38 |
+
Jinja2==3.1.2
|
| 39 |
+
jsonschema==4.17.3
|
| 40 |
+
keras==2.12.0
|
| 41 |
+
kiwisolver==1.4.4
|
| 42 |
+
libclang==16.0.0
|
| 43 |
+
linkify-it-py==2.0.2
|
| 44 |
+
Markdown==3.4.3
|
| 45 |
+
markdown-it-py==2.2.0
|
| 46 |
+
MarkupSafe==2.1.2
|
| 47 |
+
matplotlib==3.7.1
|
| 48 |
+
mdit-py-plugins==0.3.3
|
| 49 |
+
mdurl==0.1.2
|
| 50 |
+
ml-dtypes==0.1.0
|
| 51 |
+
multidict==6.0.4
|
| 52 |
+
numpy==1.23.5
|
| 53 |
+
oauthlib==3.2.2
|
| 54 |
+
opt-einsum==3.3.0
|
| 55 |
+
orjson==3.8.12
|
| 56 |
+
packaging==23.1
|
| 57 |
+
pandas==2.0.1
|
| 58 |
+
Pillow==9.5.0
|
| 59 |
+
protobuf==4.23.0
|
| 60 |
+
pyasn1==0.5.0
|
| 61 |
+
pyasn1-modules==0.3.0
|
| 62 |
+
pydantic==1.10.7
|
| 63 |
+
pydub==0.25.1
|
| 64 |
+
Pygments==2.15.1
|
| 65 |
+
pyparsing==3.0.9
|
| 66 |
+
pyrsistent==0.19.3
|
| 67 |
+
python-dateutil==2.8.2
|
| 68 |
+
python-multipart==0.0.6
|
| 69 |
+
pytz==2023.3
|
| 70 |
+
PyYAML==6.0
|
| 71 |
+
requests==2.30.0
|
| 72 |
+
requests-oauthlib==1.3.1
|
| 73 |
+
rsa==4.9
|
| 74 |
+
scipy==1.10.1
|
| 75 |
+
semantic-version==2.10.0
|
| 76 |
+
six==1.16.0
|
| 77 |
+
sniffio==1.3.0
|
| 78 |
+
starlette==0.26.1
|
| 79 |
+
tensorboard==2.12.3
|
| 80 |
+
tensorboard-data-server==0.7.0
|
| 81 |
+
tensorflow==2.12.0
|
| 82 |
+
tensorflow-estimator==2.12.0
|
| 83 |
+
tensorflow-intel==2.12.0
|
| 84 |
+
tensorflow-io-gcs-filesystem==0.31.0
|
| 85 |
+
termcolor==2.3.0
|
| 86 |
+
toolz==0.12.0
|
| 87 |
+
tqdm==4.65.0
|
| 88 |
+
typing_extensions==4.5.0
|
| 89 |
+
tzdata==2023.3
|
| 90 |
+
uc-micro-py==1.0.2
|
| 91 |
+
urllib3==1.26.15
|
| 92 |
+
uvicorn==0.22.0
|
| 93 |
+
websockets==11.0.3
|
| 94 |
+
Werkzeug==2.3.4
|
| 95 |
+
wrapt==1.14.1
|
| 96 |
+
yarl==1.9.2
|
skimlit_max_trained_model/keras_metadata.pb
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f48a02c761a858c821a9d8e1916b2364a4ae0154e98ca63ba3e8cb5be12c7fa
|
| 3 |
+
size 35641
|
skimlit_max_trained_model/saved_model.pb
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd525ae8c68fa98ba75223a708d9b948d3773af280ce68dcde070c283a4300b1
|
| 3 |
+
size 14956303
|
skimlit_max_trained_model/variables/variables.index
ADDED
|
Binary file (16.3 kB). View file
|
|
|