app.py

import gradio as gr
from transformers import AutoTokenizer, AutoConfig, pipeline
import torch
import os
from torch import nn
from torch.nn import Dropout
from transformers import XLMRobertaForSequenceClassification

HF_TOKEN = os.getenv('HF_TOKEN')
hf_writer = gr.HuggingFaceDatasetSaver(HF_TOKEN, "crowdsourced-sentiment")

# Define the CustomModel class which is predicting Both SENTIMENT POLARITY & EMOTIONS
class CustomModel(XLMRobertaForSequenceClassification):
    def __init__(self, config, num_emotion_labels):
        super(CustomModel, self).__init__(config)
        self.num_emotion_labels = num_emotion_labels
        self.dropout_emotion = nn.Dropout(config.hidden_dropout_prob)
        self.emotion_classifier = nn.Sequential(
            nn.Linear(config.hidden_size, 512),
            nn.Mish(),
            nn.Dropout(0.3),
            nn.Linear(512, num_emotion_labels)
        )
        self._init_weights(self.emotion_classifier[0])
        self._init_weights(self.emotion_classifier[3])

    def _init_weights(self, module):
        if isinstance(module, nn.Linear):
            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
            if module.bias is not None:
                module.bias.data.zero_()

    def forward(self, input_ids=None, attention_mask=None, sentiment=None, labels=None):
        outputs = self.roberta(input_ids=input_ids, attention_mask=attention_mask)
        sequence_output = outputs[0]
        if len(sequence_output.shape) != 3:
            raise ValueError(f"Expected sequence_output to have 3 dimensions, got {sequence_output.shape}")
        cls_hidden_states = sequence_output[:, 0, :]
        cls_hidden_states = self.dropout_emotion(cls_hidden_states)
        emotion_logits = self.emotion_classifier(cls_hidden_states)
        with torch.no_grad():
            cls_token_state = sequence_output[:, 0, :].unsqueeze(1)
            sentiment_logits = self.classifier(cls_token_state).squeeze(1)
        if labels is not None:
            class_weights = torch.tensor([1.0] * self.num_emotion_labels).to(labels.device)
            loss_fct = nn.BCEWithLogitsLoss(pos_weight=class_weights)
            loss = loss_fct(emotion_logits, labels)
            return {"loss": loss, "emotion_logits": emotion_logits, "sentiment_logits": sentiment_logits}
        return {"emotion_logits": emotion_logits, "sentiment_logits": sentiment_logits}

# Load the tokenizer and model from the local directory
model_dir = "gsar78/HellenicSentimentAI_v2"
tokenizer = AutoTokenizer.from_pretrained(model_dir)
config = AutoConfig.from_pretrained(model_dir)
model = CustomModel.from_pretrained(model_dir, config=config, num_emotion_labels=18)

# Function to predict sentiment and emotion
def predict(texts):
    # Tokenize the input texts
    inputs = tokenizer(texts, padding=True, truncation=True, return_tensors="pt", max_length=512)

    # Move inputs to the same device as the model
    device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
    inputs = {k: v.to(device) for k, v in inputs.items()}

    # Ensure the model is on the correct device
    model.to(device)
    model.eval()  # Set the model to evaluation mode

    # Clear any gradients
    model.zero_grad()

    # Get model predictions
    with torch.no_grad():
        outputs = model(**inputs)

    # Extract logits
    emotion_logits = outputs["emotion_logits"]
    sentiment_logits = outputs["sentiment_logits"]

    # Convert logits to probabilities
    emotion_probs = torch.sigmoid(emotion_logits)
    sentiment_probs = torch.softmax(sentiment_logits, dim=1)

    # Convert tensors to lists for easier handling
    emotion_probs_list = (emotion_probs * 100).tolist()  # Convert to %
    sentiment_probs_list = (sentiment_probs * 100).tolist()  # Convert to %

    # Define the sentiment and emotion labels
    sentiment_labels = ['negative', 'neutral', 'positive']
    emotion_labels = [
        'joy', 'trust', 'excitement', 'gratitude', 'hope', 'love', 'pride',
        'anger', 'disgust', 'fear', 'sadness', 'anxiety', 'frustration', 'guilt',
        'disappointment', 'surprise', 'anticipation', 'neutral'
    ]

    # Threshold for displaying probabilities
    threshold = 0.0

    # Map emotion probabilities to their corresponding labels
    emotion_results = [
        {label: prob for label, prob in zip(emotion_labels, emotion_probs_sample) if prob > 10.0}
        for emotion_probs_sample in emotion_probs_list
    ]

    # Map sentiment probabilities to their corresponding labels
    sentiment_results = [
        {label: prob for label, prob in zip(sentiment_labels, sentiment_probs_sample) if prob > threshold}
        for sentiment_probs_sample in sentiment_probs_list
    ]

    return emotion_results, sentiment_results

def sentiment_analysis_generate_table(text):
    sentences = text.split('|')
    emotion_results, sentiment_results = predict(sentences)

    # 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(255, 165, 0);
        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>
            <th scope="col">Emotions</th>
        </tr>
    </thead>
    <tbody>
    """
    for sentence, emotions, sentiment in zip(sentences, emotion_results, sentiment_results):
        text = sentence.strip()
        sentiment_label = max(sentiment, key=sentiment.get)
        score = f"{sentiment[sentiment_label]:.2f}%"
        
        # Determine the sentiment class
        if sentiment_label.lower() == "positive":
            sentiment_class = "positive"
        elif sentiment_label.lower() == "negative":
            sentiment_class = "negative"
        else:
            sentiment_class = "neutral"

        # Generate emotion tags
        emotion_tags = ", ".join([f"{label} ({prob:.2f}%)" for label, prob in emotions.items()])

        # Generate table rows
        html += f'<tr><td>{text}</td><td>{score}</td><td><span class="label {sentiment_class}">{sentiment_label}</span></td><td>{emotion_tags}</td></tr>'

    html += """
    </tbody>
    </table>
    </body>
    </html>
    """

    return html

if __name__ == "__main__":
    iface = gr.Interface(
        fn=sentiment_analysis_generate_table,
        inputs=gr.Textbox(placeholder="Enter sentence here..."),
        outputs=gr.HTML(),
        title="Hellenic Sentiment AI - Version 2.0",
        description="A sentiment & emotion analysis model, primarily for the Greek language.<br>"
                    "Type in some text in Greek, to classify its sentiment & emotion: positive, neutral, or negative, along with detected emotions.<br>"
                    "Multiple sentences can be classified when separated by the | character.<br>"
                    "Version 2.0 - Developed by GeoSar",
        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 από εκεί."]
        ],
        allow_flagging="manual",
        flagging_options=["Incorrect", "Ambiguous"],
        flagging_callback=hf_writer,
        examples_per_page=2,
        allow_duplication=False,
        concurrency_limit="default"
    )

    iface.launch(share=True)