Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -1,15 +1,12 @@
|
|
1 |
import streamlit as st
|
2 |
import pandas as pd
|
3 |
import torch
|
4 |
-
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
5 |
-
import random as r
|
6 |
import asyncio
|
7 |
import gradio as gr
|
|
|
8 |
|
9 |
gr.Interface.load("models/APJ23/MultiHeaded_Sentiment_Analysis_Model").launch()
|
10 |
|
11 |
-
|
12 |
-
|
13 |
tokenizer = AutoTokenizer.from_pretrained("APJ23/MultiHeaded_Sentiment_Analysis_Model", local_files_only=True)
|
14 |
model = AutoModelForSequenceClassification.from_pretrained("APJ23/MultiHeaded_Sentiment_Analysis_Model")
|
15 |
|
@@ -22,14 +19,15 @@ classes = {
|
|
22 |
5: 'Insult',
|
23 |
6: 'Identity Hate'
|
24 |
}
|
|
|
25 |
@st.cache(allow_output_mutation=True)
|
26 |
-
|
27 |
inputs = tokenizer(tweet, return_tensors="pt", padding=True, truncation=True)
|
28 |
outputs = model(**inputs)
|
29 |
predicted_class = torch.argmax(outputs.logits, dim=1)
|
30 |
predicted_prob = torch.softmax(outputs.logits, dim=1)[0][predicted_class].item()
|
31 |
return classes[predicted_class], predicted_prob
|
32 |
-
|
33 |
def create_table(predictions):
|
34 |
data = {'Tweet': [], 'Highest Toxicity Class': [], 'Probability': []}
|
35 |
for tweet, prediction in predictions.items():
|
@@ -39,15 +37,25 @@ def create_table(predictions):
|
|
39 |
df = pd.DataFrame(data)
|
40 |
return df
|
41 |
|
|
|
|
|
|
|
|
|
|
|
42 |
st.title('Toxicity Prediction App')
|
43 |
-
|
44 |
-
|
45 |
-
result = await async_prediction(tweet, model, tokenizer)
|
46 |
-
return result
|
47 |
if st.button('Predict'):
|
48 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
49 |
prediction_text = f'Prediction: {predicted_class_label} ({predicted_prob:.2f})'
|
50 |
st.write(prediction_text)
|
51 |
-
|
52 |
table = create_table(predictions)
|
53 |
st.table(table)
|
|
|
1 |
import streamlit as st
|
2 |
import pandas as pd
|
3 |
import torch
|
|
|
|
|
4 |
import asyncio
|
5 |
import gradio as gr
|
6 |
+
from transformers import pipeline, AutoTokenizer, AutoModelForSequenceClassification
|
7 |
|
8 |
gr.Interface.load("models/APJ23/MultiHeaded_Sentiment_Analysis_Model").launch()
|
9 |
|
|
|
|
|
10 |
tokenizer = AutoTokenizer.from_pretrained("APJ23/MultiHeaded_Sentiment_Analysis_Model", local_files_only=True)
|
11 |
model = AutoModelForSequenceClassification.from_pretrained("APJ23/MultiHeaded_Sentiment_Analysis_Model")
|
12 |
|
|
|
19 |
5: 'Insult',
|
20 |
6: 'Identity Hate'
|
21 |
}
|
22 |
+
|
23 |
@st.cache(allow_output_mutation=True)
|
24 |
+
def predict_toxicity(tweet, model, tokenizer):
|
25 |
inputs = tokenizer(tweet, return_tensors="pt", padding=True, truncation=True)
|
26 |
outputs = model(**inputs)
|
27 |
predicted_class = torch.argmax(outputs.logits, dim=1)
|
28 |
predicted_prob = torch.softmax(outputs.logits, dim=1)[0][predicted_class].item()
|
29 |
return classes[predicted_class], predicted_prob
|
30 |
+
|
31 |
def create_table(predictions):
|
32 |
data = {'Tweet': [], 'Highest Toxicity Class': [], 'Probability': []}
|
33 |
for tweet, prediction in predictions.items():
|
|
|
37 |
df = pd.DataFrame(data)
|
38 |
return df
|
39 |
|
40 |
+
async def run_async_prediction(tweet, model, tokenizer):
|
41 |
+
loop = asyncio.get_event_loop()
|
42 |
+
prediction = await loop.run_in_executor(None, predict_toxicity, tweet, model, tokenizer)
|
43 |
+
return prediction
|
44 |
+
|
45 |
st.title('Toxicity Prediction App')
|
46 |
+
tweet_input = st.text_input('Enter a tweet to check for toxicity')
|
47 |
+
|
|
|
|
|
48 |
if st.button('Predict'):
|
49 |
+
predictions = {tweet_input: None}
|
50 |
+
loop = asyncio.new_event_loop()
|
51 |
+
asyncio.set_event_loop(loop)
|
52 |
+
prediction = loop.run_until_complete(run_async_prediction(tweet_input, model, tokenizer))
|
53 |
+
predictions[tweet_input] = prediction
|
54 |
+
loop.close()
|
55 |
+
|
56 |
+
predicted_class_label, predicted_prob = prediction
|
57 |
prediction_text = f'Prediction: {predicted_class_label} ({predicted_prob:.2f})'
|
58 |
st.write(prediction_text)
|
59 |
+
|
60 |
table = create_table(predictions)
|
61 |
st.table(table)
|