Upload test3/eden_AlexNet_CIFAR100.py with huggingface_hub

test3/eden_AlexNet_CIFAR100.py

@@ -0,0 +1,181 @@
+import torch
+import torch.nn as nn
+import torch.optim as optim
+import torchvision
+import torchvision.transforms as transforms
+from torch.utils.data import DataLoader, TensorDataset
+from sklearn.metrics import f1_score, precision_score, recall_score
+from codecarbon import EmissionsTracker
+from thop import profile 
+import time
+import pandas as pd
+import numpy as np
+import os
+import warnings
+import copy
+from datetime import timedelta
+
+# --- Configuration ---
+MODEL_NAME = "alexnet_EDEN" 
+DATASET_NAME = "CIFAR100"
+DATA_PATH = r'C:\Users\shanm\Dataset Download\CIFAR100' 
+BATCH_SIZE = 128
+ACCUMULATION_STEPS = 4  # Effective Batch Size = 512
+EPOCHS = 15
+E_UNFREEZE = 10         
+LAMBDA_L1 = 1e-5        
+DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+SAVE_DIR = "saved_models"
+os.makedirs(SAVE_DIR, exist_ok=True)
+CSV_FILENAME = f"{MODEL_NAME}_{DATASET_NAME}_stats.csv"
+
+warnings.filterwarnings("ignore")
+os.environ["CODECARBON_LOG_LEVEL"] = "error"
+
+def main():
+    # --- Phase 1: Zero-Overhead Initialization (RAM Caching) ---
+    transform = transforms.Compose([
+        transforms.Resize(224), # AlexNet pre-trained expects 224x224
+        transforms.ToTensor(),
+        transforms.Normalize((0.5071, 0.4867, 0.4408), (0.2673, 0.2564, 0.2762)),
+    ])
+
+    print(f"[*] Caching {DATASET_NAME} to System RAM for zero-I/O overhead...")
+    try:
+        full_dataset = torchvision.datasets.CIFAR100(root=DATA_PATH, train=True, download=False, transform=transform)
+    except:
+        full_dataset = torchvision.datasets.CIFAR100(root=os.path.dirname(DATA_PATH), train=True, download=False, transform=transform)
+    
+    all_data, all_targets = [], []
+    for i, (img, target) in enumerate(full_dataset):
+        all_data.append(img)
+        all_targets.append(target)
+        if i % 10000 == 0: print(f"  Loaded {i}/50000 images...")
+    
+    cached_trainset = TensorDataset(torch.stack(all_data), torch.tensor(all_targets))
+    trainloader = DataLoader(cached_trainset, batch_size=BATCH_SIZE, shuffle=True, pin_memory=True)
+
+    # --- Model Setup ---
+    model = torchvision.models.alexnet(weights='IMAGENET1K_V1')
+    model.classifier[6] = nn.Linear(4096, 100) # 100 classes for CIFAR-100
+    
+    # 1. Calculate FLOPs on a temporary clone to avoid hook attribute errors
+    print("[*] Calculating hardware metrics (FLOPs/Params)...")
+    model_for_profile = copy.deepcopy(model).to(DEVICE)
+    dummy_input = torch.randn(1, 3, 224, 224).to(DEVICE)
+    flops, params = profile(model_for_profile, inputs=(dummy_input, ), verbose=False)
+    del model_for_profile # Clean up clone
+    
+    # 2. Freeze backbone for EDEN Phase 2 (Initial State)
+    for param in model.features.parameters():
+        param.requires_grad = False
+    
+    model.to(DEVICE)
+
+    criterion = nn.CrossEntropyLoss()
+    optimizer = optim.AdamW(model.parameters(), lr=1e-3)
+    scaler = torch.cuda.amp.GradScaler() # For Automated Mixed Precision (AMP)
+
+    results = []
+    cumulative_total_energy = 0
+    total_start_time = time.time()
+    best_acc = 0.0 
+    
+    tracker = EmissionsTracker(measure_power_secs=1, save_to_file=False, log_level='error')
+
+    print(f"\n[MODEL INFO] FLOPs: {flops/1e9:.2f} G | Parameters: {params/1e6:.2f} M | Batch Size: {BATCH_SIZE}")
+    print(f"{'='*140}")
+    print(f"{'Epoch':<6} | {'Loss':<7} | {'Acc':<7} | {'Total(J)':<9} | {'VRAM(GB)':<9} | {'EAG':<8} | {'Status'}")
+    print(f"{'-'*140}")
+
+    for epoch in range(1, EPOCHS + 1):
+        # --- EDEN Progressive Unfreezing ---
+        if epoch == E_UNFREEZE:
+            for param in model.parameters():
+                param.requires_grad = True
+            for param_group in optimizer.param_groups:
+                param_group['lr'] = 1e-5
+            status_msg = "UNFROZEN"
+        else:
+            status_msg = "FROZEN" if epoch < E_UNFREEZE else "FINE-TUNING"
+
+        model.train()
+        tracker.start()
+        epoch_start_time = time.time()
+        running_loss, all_preds, all_labels, grad_norms = 0.0, [], [], []
+        
+        optimizer.zero_grad()
+        for i, (inputs, labels) in enumerate(trainloader):
+            inputs, labels = inputs.to(DEVICE), labels.to(DEVICE)
+            
+            with torch.cuda.amp.autocast():
+                outputs = model(inputs)
+                cls_loss = criterion(outputs, labels)
+                
+                # EDEN Sparse Training Penalty (L1)
+                l1_penalty = sum(p.abs().sum() for p in model.parameters() if p.requires_grad)
+                loss = (cls_loss + LAMBDA_L1 * l1_penalty) / ACCUMULATION_STEPS
+            
+            scaler.scale(loss).backward()
+            
+            # Gradient Accumulation Logic
+            if (i + 1) % ACCUMULATION_STEPS == 0:
+                scaler.unscale_(optimizer)
+                grad_norm = torch.nn.utils.clip_grad_norm_(model.parameters(), max_norm=1.0)
+                grad_norms.append(grad_norm.item())
+                
+                scaler.step(optimizer)
+                scaler.update()
+                optimizer.zero_grad()
+            
+            running_loss += cls_loss.item()
+            _, predicted = torch.max(outputs.data, 1)
+            all_preds.extend(predicted.cpu().numpy())
+            all_labels.extend(labels.cpu().numpy())
+
+        emissions_kg = tracker.stop()
+        duration = time.time() - epoch_start_time
+        
+        # Energy Metrics (kWh -> Joules)
+        e_gpu = tracker.final_emissions_data.gpu_energy * 3600000
+        e_cpu = tracker.final_emissions_data.cpu_energy * 3600000
+        e_ram = tracker.final_emissions_data.ram_energy * 3600000
+        total_energy = e_gpu + e_cpu + e_ram
+        cumulative_total_energy += total_energy
+        
+        acc = (np.array(all_preds) == np.array(all_labels)).mean()
+        f1 = f1_score(all_labels, all_preds, average='macro')
+        vram_peak = torch.cuda.max_memory_allocated(DEVICE) / (1024**3)
+        eag = acc / (total_energy / 1000) if total_energy > 0 else 0
+
+        # CSV Logging
+        epoch_stats = {
+            "epoch": epoch, "status": status_msg, "loss": running_loss / len(trainloader),
+            "accuracy": acc, "f1_score": f1,
+            "precision": precision_score(all_labels, all_preds, average='macro', zero_division=0),
+            "recall": recall_score(all_labels, all_preds, average='macro', zero_division=0),
+            "energy_gpu_j": e_gpu, "energy_cpu_j": e_cpu, "energy_ram_j": e_ram,
+            "total_energy_j": total_energy, "cumulative_total_energy_j": cumulative_total_energy,
+            "carbon_kg": emissions_kg, "vram_gb": vram_peak,
+            "latency_ms": (duration / len(trainloader)) * 1000,
+            "eag_metric": eag, "grad_norm": np.mean(grad_norms) if grad_norms else 0,
+            "model_flops": flops, "model_params": params,
+            "batch_size": BATCH_SIZE, "accumulation_steps": ACCUMULATION_STEPS
+        }
+        results.append(epoch_stats)
+        pd.DataFrame(results).to_csv(CSV_FILENAME, index=False)
+        
+        if acc > best_acc:
+            best_acc = acc
+            torch.save(model.state_dict(), os.path.join(SAVE_DIR, f"BEST_{MODEL_NAME}_{DATASET_NAME}.pth"))
+            best_tag = "*"
+        else:
+            best_tag = ""
+
+        print(f"{epoch:02d}/50  | {epoch_stats['loss']:.4f} | {acc:.2%} | {total_energy:<9.2f} | {vram_peak:<9.3f} | {eag:<8.4f} | {status_msg}{best_tag}")
+
+    print(f"{'='*140}\n[FINISH] AlexNet on CIFAR-100 saved to {CSV_FILENAME}")
+
+if __name__ == '__main__':
+    main()