""" Example usage of the novel AI model. """ import torch import sys import os sys.path.append(os.path.dirname(os.path.abspath(__file__))) from config.config import Config from model.architecture import NovelAIModel from core.formula import ScoringFormula def example_formula_usage(): """Example of using the scoring formula directly.""" print("=" * 60) print("Example 1: Using the Scoring Formula Directly") print("=" * 60) # Create formula embedding_dim = 128 formula = ScoringFormula( embedding_dim=embedding_dim, w1=0.4, w2=0.3, w3=0.3, lambda_decay=0.1, k_fatigue=0.2 ) # Create sample data batch_size = 4 current = torch.randn(batch_size, embedding_dim) context = torch.randn(batch_size, embedding_dim) time_steps = torch.arange(batch_size, dtype=torch.float) # Compute scores scores, components = formula(current, context, time_steps=time_steps) print(f"\nInput shape: {current.shape}") print(f"Scores shape: {scores.shape}") print(f"\nSample scores: {scores[:3]}") print(f"\nComponent breakdown:") print(f" Novelty (Delta A): {components['novelty'][:3]}") print(f" Retention (R): {components['retention'][:3]}") print(f" Payoff (M): {components['payoff'][:3]}") print(f" Continuity (C): {components['continuity'][:3]}") print(f" Fatigue (phi): {components['fatigue'][:3]}") print(f"\nFormula weights:") print(f" w1 = {components['weights']['w1']:.4f}") print(f" w2 = {components['weights']['w2']:.4f}") print(f" w3 = {components['weights']['w3']:.4f}") def example_model_forward(): """Example of forward pass through the full model.""" print("\n" + "=" * 60) print("Example 2: Forward Pass Through Full Model") print("=" * 60) # Create config config = Config.default() config.model.vocab_size = 1000 config.model.embedding_dim = 128 config.model.num_layers = 2 config.model.max_seq_length = 64 # Create model formula_config = { 'w1': config.model.formula_w1, 'w2': config.model.formula_w2, 'w3': config.model.formula_w3, 'lambda_decay': config.model.formula_lambda, 'k_fatigue': config.model.formula_k, } model = NovelAIModel( vocab_size=config.model.vocab_size, embedding_dim=config.model.embedding_dim, num_layers=config.model.num_layers, max_seq_length=config.model.max_seq_length, formula_config=formula_config ) # Create sample input batch_size = 2 seq_len = 16 input_ids = torch.randint(0, config.model.vocab_size, (batch_size, seq_len)) print(f"\nInput shape: {input_ids.shape}") print(f"Model parameters: {sum(p.numel() for p in model.parameters()):,}") # Forward pass output = model(input_ids=input_ids, return_components=False) print(f"\nOutput logits shape: {output['logits'].shape}") print(f"Hidden states shape: {output['hidden_states'].shape}") # Show prediction for first token of first sequence logits = output['logits'] probs = torch.softmax(logits[0, 0, :], dim=-1) top_k = 5 top_probs, top_indices = torch.topk(probs, top_k) print(f"\nTop {top_k} predictions for first token:") for i, (idx, prob) in enumerate(zip(top_indices, top_probs)): print(f" {i+1}. Token {idx.item()}: {prob.item():.4f}") def example_training_setup(): """Example of setting up training.""" print("\n" + "=" * 60) print("Example 3: Training Setup") print("=" * 60) from utils.tokenizer import SimpleTokenizer from utils.data_loader import create_data_loaders # Sample texts train_texts = [ "This is a sample training text for the novel AI model.", "The model uses a novel scoring formula instead of traditional attention.", "Each component of the formula has a specific meaning.", "Novelty measures information gain.", "Retention estimates long-term value.", ] val_texts = [ "This is a validation text.", "It tests the model's generalization.", ] print(f"\nTraining texts: {len(train_texts)}") print(f"Validation texts: {len(val_texts)}") # Create data loaders train_loader, val_loader, tokenizer = create_data_loaders( train_texts=train_texts, val_texts=val_texts, max_length=32, batch_size=2 ) print(f"\nVocabulary size: {tokenizer.vocab_size}") print(f"Training batches: {len(train_loader)}") print(f"Validation batches: {len(val_loader)}") # Show a sample batch sample_batch = next(iter(train_loader)) print(f"\nSample batch:") print(f" Input IDs shape: {sample_batch['input_ids'].shape}") print(f" Target IDs shape: {sample_batch['target_ids'].shape}") # Decode a sample decoded = tokenizer.decode(sample_batch['input_ids'][0].tolist()) print(f"\nDecoded sample:") print(f" {decoded[:100]}...") def example_generation(): """Example of text generation.""" print("\n" + "=" * 60) print("Example 4: Text Generation") print("=" * 60) # Create a small model for demonstration config = Config.default() config.model.vocab_size = 100 config.model.embedding_dim = 64 config.model.num_layers = 2 config.model.max_seq_length = 32 formula_config = { 'w1': 0.4, 'w2': 0.3, 'w3': 0.3, 'lambda_decay': 0.1, 'k_fatigue': 0.2, } model = NovelAIModel( vocab_size=config.model.vocab_size, embedding_dim=config.model.embedding_dim, num_layers=config.model.num_layers, max_seq_length=config.model.max_seq_length, formula_config=formula_config ) model.eval() # Create a prompt prompt_ids = torch.randint(0, config.model.vocab_size, (1, 5)) print(f"\nPrompt tokens: {prompt_ids[0].tolist()}") # Generate with torch.no_grad(): generated = model.generate( input_ids=prompt_ids, max_new_tokens=10, temperature=1.0, do_sample=True ) print(f"\nGenerated tokens: {generated[0].tolist()}") print(f"Generated length: {generated.shape[1]}") if __name__ == '__main__': print("\n" + "=" * 60) print("Novel AI Model - Usage Examples") print("=" * 60) try: example_formula_usage() example_model_forward() example_training_setup() example_generation() print("\n" + "=" * 60) print("All examples completed successfully!") print("=" * 60) except Exception as e: print(f"\nError: {e}") import traceback traceback.print_exc()