Update app.py
Browse files
app.py
CHANGED
@@ -1,43 +1,20 @@
|
|
1 |
import streamlit as st
|
2 |
from transformers import pipeline
|
3 |
|
4 |
-
classifier = pipeline("
|
5 |
def main():
|
6 |
-
st.title("
|
7 |
|
8 |
-
with st.form("
|
9 |
-
|
10 |
# clicked==True only when the button is clicked
|
11 |
clicked = st.form_submit_button("Submit")
|
12 |
if clicked:
|
13 |
-
results = classifier([
|
14 |
st.json(results)
|
15 |
-
from transformers import VisionEncoderDecoderModel, ViTFeatureExtractor, AutoTokenizer
|
16 |
-
|
17 |
-
model = VisionEncoderDecoderModel.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
|
18 |
-
feature_extractor = ViTFeatureExtractor.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
|
19 |
-
tokenizer = AutoTokenizer.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
|
20 |
-
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
21 |
-
model.to(device)
|
22 |
-
max_length = 16
|
23 |
-
num_beams = 4
|
24 |
-
gen_kwargs = {"max_length": max_length, "num_beams": num_beams}
|
25 |
-
def predict_step(image_paths):
|
26 |
-
images = []
|
27 |
-
for image_path in image_paths:
|
28 |
-
i_image = Image.open(image_path)
|
29 |
-
if i_image.mode != "RGB":
|
30 |
-
i_image = i_image.convert(mode="RGB")
|
31 |
-
images.append(i_image)
|
32 |
-
pixel_values = feature_extractor(images=images, return_tensors="pt").pixel_values
|
33 |
-
pixel_values = pixel_values.to(device)
|
34 |
-
output_ids = model.generate(pixel_values, **gen_kwargs)
|
35 |
-
preds = tokenizer.batch_decode(output_ids, skip_special_tokens=True)
|
36 |
-
preds = [pred.strip() for pred in preds]
|
37 |
-
return preds
|
38 |
-
predict_step(['doctor.e16ba4e4.jpg']
|
39 |
|
40 |
if __name__ == "__main__":
|
41 |
main()
|
42 |
|
|
|
43 |
"""'audio-classification', 'automatic-speech-recognition', 'conversational', 'document-question-answering', 'feature-extraction', 'fill-mask', 'image-classification', 'image-segmentation', 'image-to-text', 'ner', 'object-detection', 'question-answering', 'sentiment-analysis', 'summarization', 'table-question-answering', 'text-classification', 'text-generation', 'text2text-generation', 'token-classification', 'translation', 'visual-question-answering', 'vqa', 'zero-shot-classification', 'zero-shot-image-classification', 'translation_XX_to_YY'"""
|
|
|
1 |
import streamlit as st
|
2 |
from transformers import pipeline
|
3 |
|
4 |
+
classifier = pipeline("token-classification", model="samrawal/medical-sentence-tokenizer")
|
5 |
def main():
|
6 |
+
st.title("Token classification")
|
7 |
|
8 |
+
with st.form("text_field"):
|
9 |
+
text = st.text_area('enter some text:')
|
10 |
# clicked==True only when the button is clicked
|
11 |
clicked = st.form_submit_button("Submit")
|
12 |
if clicked:
|
13 |
+
results = classifier([text])
|
14 |
st.json(results)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
|
16 |
if __name__ == "__main__":
|
17 |
main()
|
18 |
|
19 |
+
|
20 |
"""'audio-classification', 'automatic-speech-recognition', 'conversational', 'document-question-answering', 'feature-extraction', 'fill-mask', 'image-classification', 'image-segmentation', 'image-to-text', 'ner', 'object-detection', 'question-answering', 'sentiment-analysis', 'summarization', 'table-question-answering', 'text-classification', 'text-generation', 'text2text-generation', 'token-classification', 'translation', 'visual-question-answering', 'vqa', 'zero-shot-classification', 'zero-shot-image-classification', 'translation_XX_to_YY'"""
|