utils/prediction.py

from utils.model import BiLSTMAttentionBERT
import torch
from transformers import AutoTokenizer
from sklearn.preprocessing import LabelEncoder
import numpy as np
import streamlit as st
import requests



def load_model_for_prediction():
    try:
        st.write("Starting model loading...")
        
        # Test Hugging Face connectivity
        st.write("Testing connection to Hugging Face...")
        response = requests.get("https://huggingface.co/joko333/BiLSTM_v01")
        if response.status_code != 200:
            st.error(f"Cannot connect to Hugging Face. Status code: {response.status_code}")
            return None, None, None
            
        # Load model with logging
        st.write("Loading BiLSTM model...")
        model = BiLSTMAttentionBERT.from_pretrained(
            "joko333/BiLSTM_v01",
            hidden_dim=128,
            num_classes=22,
            num_layers=2,
            dropout=0.5
        )
        st.write("Model loaded successfully")
        
        # Initialize label encoder
        st.write("Initializing label encoder...")
        label_encoder = LabelEncoder()
        label_encoder.classes_ = np.array(['Addition', 'Causal', 'Cause and Effect', 
                                         'Clarification', 'Comparison', 'Concession', 
                                         'Conditional', 'Contrast', 'Contrastive Emphasis',
                                         'Definition', 'Elaboration', 'Emphasis', 
                                         'Enumeration', 'Explanation', 'Generalization', 
                                         'Illustration', 'Inference', 'Problem Solution', 
                                         'Purpose', 'Sequential', 'Summary', 
                                         'Temporal Sequence'])
        st.write("Label encoder initialized")
        
        # Load tokenizer
        st.write("Loading tokenizer...")
        tokenizer = AutoTokenizer.from_pretrained('dmis-lab/biobert-base-cased-v1.2')
        st.write("Tokenizer loaded successfully")
        
        return model, label_encoder, tokenizer
        
    except Exception as e:
        st.error(f"Detailed error: {str(e)}")
        st.error(f"Error type: {type(e).__name__}")
        import traceback
        st.error(f"Traceback: {traceback.format_exc()}")
        return None, None, None

def predict_sentence(model, sentence, tokenizer, label_encoder):
    """
    Make prediction for a single sentence with label validation.
    """
    # Validation checks
    if model is None:
        print("Error: Model not loaded")
        return "Error: Model not loaded", 0.0
    if tokenizer is None:
        print("Error: Tokenizer not loaded")
        return "Error: Tokenizer not loaded", 0.0
    if label_encoder is None:
        print("Error: Label encoder not loaded")
        return "Error: Label encoder not loaded", 0.0
        
    # Force CPU device
    device = torch.device('cpu')
    model = model.to(device)
    model.eval()
    
    # Tokenize
    try:
        encoding = tokenizer(
            sentence,
            add_special_tokens=True,
            max_length=512,
            padding='max_length',
            truncation=True,
            return_tensors='pt'
        ).to(device)
        
        with torch.no_grad():
            outputs = model(encoding['input_ids'], encoding['attention_mask'])
            probabilities = torch.softmax(outputs, dim=1)
            prob, pred_idx = torch.max(probabilities, dim=1)
            predicted_label = label_encoder.classes_[pred_idx.item()]
            return predicted_label, prob.item()
            
    except Exception as e:
        print(f"Prediction error: {str(e)}")
        return f"Error: {str(e)}", 0.0
    
def print_labels(label_encoder, show_counts=False):
    """Print all labels and their corresponding indices"""
    print("\nAvailable labels:")
    print("-" * 40)
    for idx, label in enumerate(label_encoder.classes_):
        print(f"Index {idx}: {label}")
    print("-" * 40)
    print(f"Total number of classes: {len(label_encoder.classes_)}\n")