see/inference.py

# -*- coding: utf-8 -*-
"""
Inference Module - Model Prediction
"""
import os
import torch

# Model path
MODEL_SAVE_PATH = '../best_model'

# Emotion labels
EMOTION_LABELS = [
    "Neutral",
    "Anxiety/Fear",
    "Anger/Frustration",
    "Sadness/Helplessness",
    "Confusion/Doubt",
    "Gratitude/Relief"
]

try:
    from transformers import AutoTokenizer, AutoModelForSequenceClassification
    MODEL_LOADED = True
except ImportError:
    MODEL_LOADED = False


class EmotionClassifier:
    """Emotion Classification Inference"""
    
    def __init__(self):
        self.tokenizer = None
        self.model = None
        self.device = None
        self.loaded = False
    
    def load_model(self, model_path=None):
        """Load model"""
        if model_path is None:
            model_path = MODEL_SAVE_PATH
        
        if not MODEL_LOADED:
            return {'error': 'transformers library not installed'}
        
        try:
            self.tokenizer = AutoTokenizer.from_pretrained(model_path)
            self.model = AutoModelForSequenceClassification.from_pretrained(model_path)
            self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
            self.model.to(self.device)
            self.model.eval()
            self.loaded = True
            
            return {
                'success': True,
                'device': str(self.device),
                'num_labels': len(EMOTION_LABELS),
                'labels': EMOTION_LABELS
            }
        except Exception as e:
            return {'error': f'Failed to load model: {str(e)}'}
    
    def predict(self, text, max_length=512):
        """Predict emotion for single text"""
        if not self.loaded:
            result = self.load_model()
            if 'error' in result:
                return result
        
        try:
            # Tokenize
            inputs = self.tokenizer(
                text, 
                return_tensors="pt", 
                padding=True, 
                truncation=True, 
                max_length=max_length
            )
            
            # Move to device
            inputs = {k: v.to(self.device) for k, v in inputs.items()}
            
            # Inference
            with torch.no_grad():
                outputs = self.model(**inputs)
                logits = outputs.logits
                probabilities = torch.softmax(logits, dim=-1)
                predicted_class = torch.argmax(logits, dim=-1).item()
                confidence = probabilities[0][predicted_class].item()
            
            # Build result
            all_probs = probabilities[0].cpu().numpy().tolist()
            label_probs = [
                {'label': EMOTION_LABELS[i], 'probability': round(all_probs[i], 4)}
                for i in range(len(EMOTION_LABELS))
            ]
            
            return {
                'text': text[:100] + '...' if len(text) > 100 else text,
                'predicted_label': EMOTION_LABELS[predicted_class],
                'predicted_id': predicted_class,
                'confidence': round(confidence, 4),
                'all_probabilities': label_probs
            }
            
        except Exception as e:
            return {'error': f'Prediction failed: {str(e)}'}
    
    def predict_batch(self, texts, max_length=512):
        """Batch prediction"""
        if not self.loaded:
            result = self.load_model()
            if 'error' in result:
                return result
        
        results = []
        for text in texts:
            result = self.predict(text, max_length)
            results.append(result)
        
        return results
    
    def is_loaded(self):
        """Check if model is loaded"""
        return self.loaded


# Global classifier instance
_classifier_instance = None

def get_classifier():
    global _classifier_instance
    if _classifier_instance is None:
        _classifier_instance = EmotionClassifier()
    return _classifier_instance