Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
| 3 |
+
import torch
|
| 4 |
+
|
| 5 |
+
# Load your model and tokenizer
|
| 6 |
+
model_name = "your-hf-username/LexCAT-LexiLiksik-Final" # ← REPLACE WITH YOUR MODEL REPO
|
| 7 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
| 8 |
+
model = AutoModelForSequenceClassification.from_pretrained(model_name)
|
| 9 |
+
|
| 10 |
+
# Define prediction function
|
| 11 |
+
def predict_sentiment(text):
|
| 12 |
+
if not text.strip():
|
| 13 |
+
return "Please enter text."
|
| 14 |
+
|
| 15 |
+
inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=128)
|
| 16 |
+
outputs = model(**inputs)
|
| 17 |
+
predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)
|
| 18 |
+
predicted_class = torch.argmax(predictions, dim=1).item()
|
| 19 |
+
|
| 20 |
+
sentiment_map = {0: "❌ Negative", 1: "➖ Neutral", 2: "✅ Positive"}
|
| 21 |
+
confidence = predictions[0][predicted_class].item()
|
| 22 |
+
|
| 23 |
+
result = f"**Predicted Sentiment**: {sentiment_map[predicted_class]}\n"
|
| 24 |
+
result += f"**Confidence**: {confidence:.3f}\n"
|
| 25 |
+
result += f"\n**Raw Probabilities**:\nNegative: {predictions[0][0]:.3f}\nNeutral: {predictions[0][1]:.3f}\nPositive: {predictions[0][2]:.3f}"
|
| 26 |
+
|
| 27 |
+
return result
|
| 28 |
+
|
| 29 |
+
# Define Gradio interface
|
| 30 |
+
demo = gr.Interface(
|
| 31 |
+
fn=predict_sentiment,
|
| 32 |
+
inputs=gr.Textbox(placeholder="Type a Taglish sentence, e.g., 'Maganda pero expensive tlga'", label="Input Tagalog–English (Taglish) Text"),
|
| 33 |
+
outputs=gr.Textbox(label="Sentiment Prediction"),
|
| 34 |
+
title="🔍 LexCAT: Taglish Sentiment Analysis",
|
| 35 |
+
description="""
|
| 36 |
+
**LexCAT** is a lexicon-enhanced transformer model for sentiment analysis of **Tagalog–English code-switched text (Taglish)**.
|
| 37 |
+
Developed by Glenn Marcus D. Cinco for his BS/MS thesis at Mapúa University.
|
| 38 |
+
Trained on FiReCS dataset, enhanced with **LexiLiksik** lexicon for intra-sentential shift detection (e.g., “Maganda pero expensive” → ❌ Negative).
|
| 39 |
+
""",
|
| 40 |
+
examples=[
|
| 41 |
+
["sobrang lambot ng burger pero expensive tlga"],
|
| 42 |
+
["Ang ganda ng service, one star!"],
|
| 43 |
+
["Super duper late delivery umabot ng 2 weeks metro manila area lang naman"],
|
| 44 |
+
["Salamat sa nyo nagana nmn po sya kaya super thank you ako"],
|
| 45 |
+
["Ganda legit, kumpleto... problema lang nainit ng sobra..."]
|
| 46 |
+
],
|
| 47 |
+
theme="soft",
|
| 48 |
+
allow_flagging="never"
|
| 49 |
+
)
|
| 50 |
+
|
| 51 |
+
# Launch app
|
| 52 |
+
if __name__ == "__main__":
|
| 53 |
+
demo.launch()
|