| ---
|
| language: en
|
| license: apache-2.0
|
| tags:
|
| - image-classification
|
| - green-ai
|
| - energy-efficiency
|
| - computer-vision
|
| - alexnet
|
| - eden-framework
|
| - e2am
|
| - sustainable-ai
|
| datasets:
|
| - cifar100
|
| metrics:
|
| - accuracy
|
| co2_eq_emissions:
|
| emissions: 3.6739
|
| unit: kg
|
| source: Estimated via CodeCarbon (grid factor 0.475 kg CO2e/kWh)
|
| hardware_used: NVIDIA GeForce GTX 1080 Ti
|
| dataset_info:
|
| dataset_size: "60,000 images – 100 classes (32×32 px)"
|
| model-index:
|
| - name: EDEN-AlexNet-CIFAR-100
|
| results:
|
| - task:
|
| type: image-classification
|
| name: Image Classification
|
| dataset:
|
| name: CIFAR-100
|
| type: cifar100
|
| metrics:
|
| - type: accuracy
|
| value: 0.8933
|
| name: Accuracy
|
| - type: f1
|
| value: 0.8932
|
| name: F1 Score
|
| ---
|
|
|
| # EDEN-AlexNet-CIFAR-100 — *EDEN Classic*
|
|
|
| > **Primary KPI:** EAG (Energy-to-Accuracy Gradient) = `4.5587e-09` ΔAcc/ΔJoules
|
|
|
| ## Abstract
|
| This model is part of **Project EDEN (Energy-Driven Evolution of Networks)**, implementing the
|
| **E2AM (Energy Efficient Advanced Model)** Framework. The goal is to shift AI benchmarking from
|
| pure accuracy to *Green SOTA* — maximising predictive power per Joule consumed.
|
|
|
| **Applied Technique:** Phase 2 – EDEN Classic Energy-Aware Sparse Training
|
|
|
| ## Profiling Environment
|
| | Component | Specification |
|
| |---|---|
|
| | **GPU** | NVIDIA GeForce GTX 1080 Ti (11 GB VRAM, 250 W TDP) |
|
| | **CPU** | Intel Xeon W-2125 (4 cores / 8 threads @ 4.00 GHz) |
|
| | **RAM** | 63.66 GB System RAM |
|
| | **OS** | Windows 10 |
|
| | **Dataset** | CIFAR-100 — 60,000 images – 100 classes (32×32 px) |
|
|
|
| ## 🟢 Green Delta Table
|
| *Comparing this model against the reference baseline (ResNet-50 equivalent)*
|
|
|
| | Metric | ResNet50 Baseline | **AlexNet (EDEN)** | Δ |
|
| |---|---|---|---|
|
| | Accuracy | 0.9492 | **0.8933** | `-5.59%` |
|
| | Total Energy (J) | 40,102,666 | **27,844,651** | `30.57% saved` |
|
| | CO₂ Emissions (kg) | 5.2913 | **3.6739** | — |
|
| | **EAG Score** | — | **4.5587e-09** | ΔAcc/ΔJoules |
|
|
|
| > A **positive EAG** means this model learns more per Joule than the baseline.
|
| > A **negative EAG** indicates a trade-off where higher accuracy required more energy investment.
|
|
|
| ## E2AM Algorithm — Applied Phases
|
|
|
| **Phase 1 – Zero-Overhead Initialization:** Dataset cached in System RAM.
|
|
|
| **Phase 2 – EDEN Classic:** Energy-aware training loop on classic CNN architectures. Applies the same EAG early-exit criterion (`EAG < γ_EAG` for 3 consecutive epochs → terminate), L1 sparsity penalty, and AMP to architectures like ResNet, VGG, AlexNet, DenseNet, InceptionV3, and UNet.
|
|
|
| ## Training Statistics
|
| | Metric | Value |
|
| |---|---|
|
| | Final Accuracy | 0.8933 (89.33%) |
|
| | Total Energy Consumed | 27,844,651 J (7.7346 kWh) |
|
| | Training Time | 0 s (0.00 hrs) |
|
| | Estimated CO₂ | 3.6739 kg CO₂e |
|
| | Training Log | `test3\alexnet_EDEN_CIFAR100_stats.csv` |
|
|
|
| ## 📊 Training Visualizations
|
|
|
| ### Accuracy & Energy over Training
|
| > Green = accuracy (left axis) · Orange dashed = cumulative energy (right axis)
|
|
|
| 
|
|
|
| ### EAG Metric Trajectory
|
| > EAG = ΔAccuracy / ΔJoules — positive means learning more per Joule than baseline
|
|
|
| 
|
|
|
| ### Project-Wide Overview
|
| *All EDEN models: energy vs accuracy*
|
|
|
| 
|
|
|
| ## Cite This Research
|
| ```bibtex
|
| @misc{eden2025,
|
| title = {Project EDEN: Energy-Driven Evolution of Networks},
|
| author = {EDEN Research Team},
|
| year = {2025},
|
| note = {Hugging Face: Shanmuk4622},
|
| url = {https://huggingface.co/Shanmuk4622}
|
| }
|
| ```
|
|
|
