Spaces:

gsar78
/

Greek_Sentiment_Analysis

Sleeping

App Files Files Community

gsar78 commited on Jul 8, 2024

Commit

6050aa2

verified ·

1 Parent(s): 2ed3a46

Update app.py

Browse files

Files changed (1) hide show

app.py +112 -47

app.py CHANGED Viewed

@@ -1,50 +1,115 @@
 import gradio as gr
-def predict(text):
-    # Tokenize the input text
-    inputs = tokenizer(text, return_tensors="pt")
-    # Get the model outputs
-    outputs = model(**inputs)
-    # Apply softmax to the logits to get probabilities
-    scores = torch.nn.functional.softmax(outputs.logits, dim=1)
-    # Get the predicted label
-    predicted_label_idx = scores.argmax(dim=1).item()
-    labels = ["negative", "neutral", "positive"]
-    predicted_label = labels[predicted_label_idx]
-    confidence_score = scores[0, predicted_label_idx].item()
-    # Create a dictionary with the prediction and scores
-    result = {
-        "text": text,
-        "label": predicted_label,
-        "score": confidence_score,
-        "scores": {
-            "positive": scores[0, 2].item(),
-            "neutral": scores[0, 1].item(),
-            "negative": scores[0, 0].item()
-        }
     }
-    return result
-iface = gr.Interface(
-    fn=predict,
-    inputs="text",
-    outputs=[
-        gr.Textbox(label="Prediction"),
-        gr.Label(label="Label Confidence")
-    ],
-    title="Hellenic Sentiment AI",
-    description=None,
-    article=None,
-    theme="default",
-    flagging_dir=None,
-    css=None,
-    analytics_script=None,
-    allow_flagging="never",
-    allow_screenshot=True,
-    enable_queue=True,
-    show_input=True,
-    show_output=True,
-    footer="Development by Geo Sar"
-)
-iface.launch(share=True)

 import gradio as gr
+from transformers import pipeline
+def sentiment_analysis_generate_text(text):
+    # Define the model
+    model_name = "gsar78/HellenicSentimentAI"
+    # Create the pipeline
+    nlp = pipeline("sentiment-analysis", model=model_name)
+    # Split the input text into individual sentences
+    sentences = text.split('|')
+    # Run the pipeline on each sentence and collect the results
+    results = nlp(sentences)
+    output = []
+    for sentence, result in zip(sentences, results):
+        output.append(f"Text: {sentence.strip()}\nSentiment: {result['label']}, Score: {result['score']:.4f}\n")
+    # Join the results into a single string to return
+    return "\n".join(output)
+def sentiment_analysis_generate_table(text):
+    # Define the model
+    model_name = "gsar78/HellenicSentimentAI"
+    # Create the pipeline
+    nlp = pipeline("sentiment-analysis", model=model_name)
+    # Split the input text into individual sentences
+    sentences = text.split('|')
+    # Generate the HTML table with enhanced colors and bold headers
+    html = """
+    <html>
+    <head>
+    <link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/[email protected]/css/bootstrap.min.css">
+    <style>
+    .label {
+        transition: .15s;
+        border-radius: 8px;
+        padding: 5px 10px;
+        font-size: 14px;
+        text-transform: uppercase;
+    }
+    .positive {
+        background-color: rgb(54, 176, 75);
+        color: white;
+    }
+    .negative {
+        background-color: rgb(237, 83, 80);
+        color: white;
+    }
+    .neutral {
+        background-color: rgb(52, 152, 219);
+        color: white;
     }
+    th {
+        font-weight: bold;
+        color: rgb(106, 38, 198);
+    }
+    </style>
+    </head>
+    <body>
+    <table class="table table-striped">
+    <thead>
+        <tr>
+            <th scope="col">Text</th>
+            <th scope="col">Score</th>
+            <th scope="col">Sentiment</th>
+        </tr>
+    </thead>
+    <tbody>
+    """
+    for sentence in sentences:
+        result = nlp(sentence.strip())[0]
+        text = sentence.strip()
+        score = f"{result['score']:.4f}"
+        sentiment = result['label']
+        # Determine the sentiment class
+        if sentiment == "Positive":
+            sentiment_class = "positive"
+        elif sentiment == "Negative":
+            sentiment_class = "negative"
+        else:
+            sentiment_class = "neutral"
+        # Generate table rows
+        html += f'<tr><td>{text}</td><td>{score}</td><td><span class="label {sentiment_class}">{sentiment}</span></td></tr>'
+    html += """
+    </tbody>
+    </table>
+    </body>
+    </html>
+    """
+    return html
+if __name__ == "__main__":
+    iface = gr.Interface(
+        sentiment_analysis_generate_table,
+        gr.Textbox(placeholder="Enter sentence here..."),
+        ["html"],
+        title="Hellenic Sentiment AI",
+        description="<p>A sentiment analysis model for primarily the Greek language</p>"
+                    "<p>Enter some text to see whether the sentiment is positive, neutral or negative.</p>",
+        examples=[
+            ['Η πικάντικη γεύση αυτής της σούπας λαχανικών ήταν ακριβώς αυτό που χρειαζόμουν σήμερα. Είχε μια ωραία γαργαλιστική αίσθηση χωρίς να είναι πολύ καυτερή.'],
+            ['Η πίτσα ήταν καμένη και τα υλικά φθηνής ποιότητας. Σίγουρα δεν θα ξαναπαραγγείλω απ�� εκεί.']
+        ],
+        allow_flagging=False,
+        examples_per_page=2,
+    )
+    iface.launch()