Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,43 +1,31 @@
|
|
1 |
import streamlit as st
|
2 |
-
import
|
3 |
-
import
|
4 |
-
from transformers import pipeline, AutoTokenizer
|
5 |
|
6 |
-
|
7 |
-
|
8 |
-
st.sidebar.title("Sentiment Analysis App")
|
9 |
-
st.sidebar.write("Choose a pre-trained model for sentiment analysis")
|
10 |
-
model_name = st.sidebar.selectbox("Select Model", ["bert-base-uncased", "distilbert-base-uncased", "roberta-base"])
|
11 |
-
text_input = st.text_input("Enter text for sentiment analysis")
|
12 |
|
13 |
-
|
14 |
-
if text_input and isinstance(text_input, str):
|
15 |
-
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
16 |
-
encoded_input = tokenizer(text_input, return_tensors='pt', padding=True, truncation=True)
|
17 |
-
sentiment_classifier = pipeline("sentiment-analysis", model=model_name)
|
18 |
-
result = sentiment_classifier(encoded_input)[0]
|
19 |
-
st.write(f"Sentiment: {result['label']}")
|
20 |
-
st.write(f"Score: {round(result['score'], 4)}")
|
21 |
-
|
22 |
-
# create a dataframe to hold the sentiment distribution
|
23 |
-
df_sentiments = pd.DataFrame({"sentiment": ["Positive", "Negative", "Neutral"], "count": [0, 0, 0]})
|
24 |
-
|
25 |
-
# get the sentiment distribution of the text using the selected model
|
26 |
-
results = sentiment_classifier(encoded_input)
|
27 |
-
for r in results:
|
28 |
-
if r["label"] == "POSITIVE":
|
29 |
-
df_sentiments.loc[0, "count"] = r["score"]
|
30 |
-
elif r["label"] == "NEGATIVE":
|
31 |
-
df_sentiments.loc[1, "count"] = r["score"]
|
32 |
-
elif r["label"] == "NEUTRAL":
|
33 |
-
df_sentiments.loc[2, "count"] = r["score"]
|
34 |
-
|
35 |
-
# plot the sentiment distribution as a pie chart
|
36 |
-
fig = px.pie(df_sentiments, values='count', names='sentiment', hole=.4)
|
37 |
-
st.plotly_chart(fig)
|
38 |
-
|
39 |
-
else:
|
40 |
-
st.warning("Please enter some valid text for sentiment analysis")
|
41 |
|
42 |
-
|
43 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import streamlit as st
|
2 |
+
import torch
|
3 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
|
|
4 |
|
5 |
+
tokenizer = AutoTokenizer.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
|
6 |
+
model = AutoModelForSequenceClassification.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
|
|
|
|
|
|
|
|
|
7 |
|
8 |
+
st.title("Sentiment Analysis App")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
|
10 |
+
text = st.text_input("Enter text to analyze:")
|
11 |
+
if st.button("Analyze"):
|
12 |
+
encoding = tokenizer.encode_plus(text, return_tensors="pt", padding=True, truncation=True)
|
13 |
+
input_ids = encoding["input_ids"]
|
14 |
+
attention_mask = encoding["attention_mask"]
|
15 |
+
|
16 |
+
with torch.no_grad():
|
17 |
+
output = model(input_ids, attention_mask)
|
18 |
+
prediction = int(torch.argmax(output.logits))
|
19 |
+
|
20 |
+
if prediction == 0:
|
21 |
+
st.write("Negative")
|
22 |
+
elif prediction == 1:
|
23 |
+
st.write("Neutral")
|
24 |
+
else:
|
25 |
+
st.write("Positive")
|
26 |
+
|
27 |
+
values = [output.logits[0][0].item(), output.logits[0][1].item(), output.logits[0][2].item()]
|
28 |
+
labels = ["Negative", "Neutral", "Positive"]
|
29 |
+
fig, ax = plt.subplots()
|
30 |
+
ax.bar(labels, values)
|
31 |
+
st.pyplot(fig)
|