Update app.py
Browse files
app.py
CHANGED
|
@@ -1,17 +1,8 @@
|
|
|
|
|
| 1 |
from pptx import Presentation
|
| 2 |
import re
|
| 3 |
-
import gradio as gr
|
| 4 |
-
from transformers import AutoModelForSequenceClassification, AutoTokenizer
|
| 5 |
-
import torch
|
| 6 |
-
import torch.nn.functional as F
|
| 7 |
from transformers import pipeline
|
| 8 |
|
| 9 |
-
# Load the pre-trained model and tokenizer
|
| 10 |
-
tokenizer = AutoTokenizer.from_pretrained("Ahmed235/roberta_classification")
|
| 11 |
-
model = AutoModelForSequenceClassification.from_pretrained("Ahmed235/roberta_classification")
|
| 12 |
-
device = torch.device("cpu")
|
| 13 |
-
model = model.to(device) # Move the model to the CPU
|
| 14 |
-
|
| 15 |
# Create a summarization pipeline
|
| 16 |
summarizer = pipeline("summarization", model="Falconsai/text_summarization")
|
| 17 |
|
|
@@ -26,30 +17,13 @@ def extract_text_from_pptx(file_path):
|
|
| 26 |
|
| 27 |
def predict_pptx_content(file_path):
|
| 28 |
try:
|
| 29 |
-
|
| 30 |
-
|
| 31 |
-
|
| 32 |
-
# Tokenize and encode the cleaned text
|
| 33 |
-
input_encoding = tokenizer(cleaned_text, truncation=True, padding=True, return_tensors="pt")
|
| 34 |
-
input_encoding = {key: val.to(device) for key, val in input_encoding.items()} # Move input tensor to CPU
|
| 35 |
-
|
| 36 |
-
# Perform inference
|
| 37 |
-
with torch.no_grad():
|
| 38 |
-
outputs = model(**input_encoding)
|
| 39 |
-
logits = outputs.logits
|
| 40 |
-
|
| 41 |
-
probabilities = F.softmax(logits, dim=1)
|
| 42 |
-
|
| 43 |
-
predicted_label_id = torch.argmax(logits, dim=1).item()
|
| 44 |
-
predicted_label = model.config.id2label[predicted_label_id]
|
| 45 |
-
predicted_probability = probabilities[0][predicted_label_id].item()
|
| 46 |
|
| 47 |
# Summarize the cleaned text
|
| 48 |
summary = summarizer(cleaned_text, max_length=80, min_length=30, do_sample=False)[0]['summary_text']
|
| 49 |
|
| 50 |
prediction = {
|
| 51 |
-
"Predicted Label": predicted_label,
|
| 52 |
-
"Evaluation": f"Evaluate the topic according to {predicted_label} is: {predicted_probability}",
|
| 53 |
"Summary": summary
|
| 54 |
}
|
| 55 |
|
|
@@ -64,10 +38,10 @@ def predict_pptx_content(file_path):
|
|
| 64 |
iface = gr.Interface(
|
| 65 |
fn=predict_pptx_content,
|
| 66 |
inputs=gr.File(type="filepath", label="Upload PowerPoint (.pptx) file"),
|
| 67 |
-
outputs=
|
| 68 |
live=False, # Change to True for one-time analysis
|
| 69 |
title="<h1 style='color: lightgreen; text-align: center;'>HackTalk Analyzer</h1>",
|
| 70 |
)
|
| 71 |
|
| 72 |
# Deploy the Gradio interface
|
| 73 |
-
iface.launch(share=True)
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
from pptx import Presentation
|
| 3 |
import re
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
from transformers import pipeline
|
| 5 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
# Create a summarization pipeline
|
| 7 |
summarizer = pipeline("summarization", model="Falconsai/text_summarization")
|
| 8 |
|
|
|
|
| 17 |
|
| 18 |
def predict_pptx_content(file_path):
|
| 19 |
try:
|
| 20 |
+
extracted_text = extract_text_from_pptx(file_path)
|
| 21 |
+
cleaned_text = re.sub(r'\s+', ' ', extracted_text)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
|
| 23 |
# Summarize the cleaned text
|
| 24 |
summary = summarizer(cleaned_text, max_length=80, min_length=30, do_sample=False)[0]['summary_text']
|
| 25 |
|
| 26 |
prediction = {
|
|
|
|
|
|
|
| 27 |
"Summary": summary
|
| 28 |
}
|
| 29 |
|
|
|
|
| 38 |
iface = gr.Interface(
|
| 39 |
fn=predict_pptx_content,
|
| 40 |
inputs=gr.File(type="filepath", label="Upload PowerPoint (.pptx) file"),
|
| 41 |
+
outputs="text", # Only output the summary
|
| 42 |
live=False, # Change to True for one-time analysis
|
| 43 |
title="<h1 style='color: lightgreen; text-align: center;'>HackTalk Analyzer</h1>",
|
| 44 |
)
|
| 45 |
|
| 46 |
# Deploy the Gradio interface
|
| 47 |
+
iface.launch(share=True)
|