| ---
|
| language: en
|
| license: apache-2.0
|
| tags:
|
| - green-ai
|
| - energy-efficiency
|
| - e2am
|
| - eden-framework
|
| - sustainable-ai
|
| - image-classification
|
| ---
|
|
|
| # EDEN-Core-Scripts — E2AM Framework Repository
|
|
|
| > **Project EDEN (Energy-Driven Evolution of Networks)** — The complete algorithmic
|
| > toolkit for Green SOTA image classification research.
|
|
|
| ## Why EDEN?
|
| As deep learning models scale exponentially, the carbon footprint of training has reached
|
| unsustainable levels. Project EDEN introduces the **EAG (Energy-to-Accuracy Gradient)** as
|
| the primary KPI — shifting the paradigm from chasing raw accuracy to optimising *Green SOTA*.
|
|
|
| ## 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 |
|
|
|
| ---
|
|
|
| ## 📊 Collection Overview
|
|
|
| ### Energy vs Accuracy — All Models
|
| *SOTA Optimized (green) · Baseline (grey) · EDEN Classic (blue)*
|
|
|
| 
|
|
|
| ### EAG Leaderboard — Ranked by Green Efficiency
|
| 
|
|
|
| ### CO₂ Emissions — Baseline vs EDEN Classic
|
| 
|
|
|
| ---
|
|
|
| ## The E2AM Algorithm
|
|
|
| ### Phase 1 — Zero-Overhead Initialization
|
| Dataset pre-loaded into **pinned System RAM** before training — eliminates disk I/O power spikes.
|
|
|
| ### Phase 2 — Two-Stage Energy-Aware Training
|
| 1. **Frozen Head Training** — Only the classification head trains for `E_unfreeze` epochs.
|
| 2. **Progressive Unfreezing** — All layers unlock at `E_unfreeze`; LR decayed (`×0.1`).
|
| 3. **Gradient Accumulation** — Simulates large batch sizes without VRAM spikes.
|
| 4. **AMP** — `torch.cuda.amp.autocast()` halves bandwidth per backward pass.
|
| 5. **Sparse L1 Penalty** — `L_total = CrossEntropy + λ·Σ|W_trainable|`
|
| 6. **EAG Early-Exit** — Terminates if `EAG < γ_EAG` for 3 consecutive epochs.
|
|
|
| ### Phase 3 — Hardware-Aware Deployment *(Post-Training)*
|
| Saliency-energy pruning · INT8 quantization · Dynamic depth routing
|
|
|
| ## EAG — The Expert KPI
|
| ```
|
| EAG = ΔAccuracy / ΔJoules
|
| ```
|
| A higher EAG = more learning per unit of carbon footprint.
|
|
|
| ## Scripts in This Repository
|
| - `eden_chart_push.py`
|
| - `eden_check_hf.py`
|
| - `eden_fix_missing_repos.py`
|
| - `eden_hf_upload.py`
|
| - `eden_upload_fast.py`
|
| - `eden_upload_weights.py`
|
| - `test1\Algo_CIFAR_100_EfficientNet.py`
|
| - `test1\Algo_CIFAR_100_MobileViTv3.py`
|
| - `test1\Algo_CIFAR_100_convneXt.py`
|
| - `test1\Algo_CIFAR_10_EfficientNet.py`
|
| - `test1\Algo_CIFAR_10_MobileViTv3.py`
|
| - `test1\Algo_CIFAR_10_convneXt.py`
|
| - `test1\Algo_ImageNet_EfficientNet.py`
|
| - `test1\Algo_ImageNet_convnext.py`
|
| - `test1\Algo_ImageNet_mobilevit3.py`
|
| - `test1\mobilevit_model.py`
|
| - `test3\eden_AlexNet_CIFAR10.py`
|
| - `test3\eden_AlexNet_CIFAR100.py`
|
| - `test3\eden_AlexNet_ImageNet.py`
|
| - `test3\eden_DenseNet_121_CIFAR10.py`
|
| - `test3\eden_DenseNet_121_CIFAR100.py`
|
| - `test3\eden_DenseNet_121_ImageNet.py`
|
| - `test3\eden_InceptionV3_CIFAR10.py`
|
| - `test3\eden_InceptionV3_CIFAR100.py`
|
| - `test3\eden_InceptionV3_ImageNet.py`
|
| - `test3\eden_ResNet18_CIFAR10.py`
|
| - `test3\eden_ResNet18_CIFAR100.py`
|
| - `test3\eden_ResNet18_ImageNet.py`
|
| - `test3\eden_ResNet50_CIFAR10.py`
|
| - `test3\eden_ResNet50_CIFAR100.py`
|
| - `test3\eden_ResNet50_ImageNet.py`
|
| - `test3\eden_UNet_CIFAR10.py`
|
| - `test3\eden_UNet_CIFAR100.py`
|
| - `test3\eden_UNet_ImageNet.py`
|
| - `test3\eden_VGG16_CIFAR10.py`
|
| - `test3\eden_VGG16_CIFAR100.py`
|
| - `test3\eden_VGG16_ImageNet.py`
|
|
|
| ## 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}
|
| }
|
| ```
|
|
|
