Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import torch
|
3 |
+
from transformers import GPT2Tokenizer
|
4 |
+
import pandas as pd
|
5 |
+
|
6 |
+
# Load the tokenizer
|
7 |
+
tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
|
8 |
+
|
9 |
+
# Define the classification function
|
10 |
+
def classify_review(text, model, tokenizer, device, max_length=None, pad_token_id=50256):
|
11 |
+
model.eval()
|
12 |
+
|
13 |
+
# Prepare inputs to the model
|
14 |
+
input_ids = tokenizer.encode(text)
|
15 |
+
supported_context_length = model.pos_emb.weight.shape[1]
|
16 |
+
|
17 |
+
# Truncate sequences if they are too long
|
18 |
+
input_ids = input_ids[:min(max_length, supported_context_length)]
|
19 |
+
|
20 |
+
# Pad sequences to the longest sequence
|
21 |
+
input_ids += [pad_token_id] * (max_length - len(input_ids))
|
22 |
+
input_tensor = torch.tensor(input_ids, device=device).unsqueeze(0) # add batch dimension
|
23 |
+
|
24 |
+
# Model inference
|
25 |
+
with torch.no_grad():
|
26 |
+
logits = model(input_tensor)[:, -1, :] # Logits of the last output token
|
27 |
+
predicted_label = torch.argmax(logits, dim=-1).item()
|
28 |
+
|
29 |
+
# Return the classified result
|
30 |
+
return "Proper Naming Notfcn" if predicted_label == 1 else "Wrong Naming Notificn"
|
31 |
+
|
32 |
+
# Load the trained model from the local directory
|
33 |
+
model_path = "clv__classifier_774M.pth"
|
34 |
+
model = torch.load(model_path)
|
35 |
+
model.eval()
|
36 |
+
|
37 |
+
# Set the device to run the model on (GPU if available, else CPU)
|
38 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
39 |
+
model.to(device)
|
40 |
+
|
41 |
+
# Streamlit app
|
42 |
+
def main():
|
43 |
+
st.title("Text Classification App")
|
44 |
+
|
45 |
+
# Input options
|
46 |
+
input_option = st.radio("Select input option", ("Single Text Query", "Upload Table"))
|
47 |
+
|
48 |
+
if input_option == "Single Text Query":
|
49 |
+
# Single text query input
|
50 |
+
text_query = st.text_input("Enter text query")
|
51 |
+
if st.button("Classify"):
|
52 |
+
if text_query:
|
53 |
+
# Classify the text query
|
54 |
+
predicted_label = classify_review(text_query, model, tokenizer, device, max_length=train_dataset.max_length)
|
55 |
+
st.write("Predicted Label:")
|
56 |
+
st.write(predicted_label)
|
57 |
+
else:
|
58 |
+
st.warning("Please enter a text query.")
|
59 |
+
|
60 |
+
elif input_option == "Upload Table":
|
61 |
+
# Table upload
|
62 |
+
uploaded_file = st.file_uploader("Choose a file", type=["csv", "xlsx"])
|
63 |
+
if uploaded_file is not None:
|
64 |
+
# Read the uploaded file
|
65 |
+
if uploaded_file.name.endswith(".csv"):
|
66 |
+
df = pd.read_csv(uploaded_file)
|
67 |
+
else:
|
68 |
+
df = pd.read_excel(uploaded_file)
|
69 |
+
|
70 |
+
# Select the text column
|
71 |
+
text_column = st.selectbox("Select the text column", df.columns)
|
72 |
+
|
73 |
+
# Classify the texts in the selected column
|
74 |
+
predicted_labels = []
|
75 |
+
for text in df[text_column]:
|
76 |
+
predicted_label = classify_review(text, model, tokenizer, device, max_length=train_dataset.max_length)
|
77 |
+
predicted_labels.append(predicted_label)
|
78 |
+
|
79 |
+
# Add the predicted labels to the DataFrame
|
80 |
+
df["Predicted Label"] = predicted_labels
|
81 |
+
|
82 |
+
# Display the DataFrame with predicted labels
|
83 |
+
st.write(df)
|
84 |
+
|
85 |
+
if __name__ == "__main__":
|
86 |
+
main()
|