analyze_model.py

import torch
from transformers import AutoModel, AutoTokenizer
import matplotlib.pyplot as plt

def analyze_model(model_path):
    model = AutoModel.from_pretrained(model_path)
    tokenizer = AutoTokenizer.from_pretrained(model_path)
    
    print("=== Model Architecture ===")
    print(f"Model parameters: {sum(p.numel() for p in model.parameters()):,}")
    print(f"Layers: {len(model.encoder.layer) if hasattr(model, 'encoder') else 'N/A'}")
    
    # Analyze attention patterns
    if hasattr(model, 'encoder'):
        layer = model.encoder.layer[0]
        print(f"Attention heads: {layer.attention.self.num_attention_heads}")
    
    return model, tokenizer

def plot_training_metrics(log_file='training.log'):
    # Parse training logs and create plots
    # This would read your training logs and create nice graphs
    pass