saurabh-singh-rajput/green-tea-llama-3.1-8b-energy-sft
saurabh-singh-rajput/green-tea-llama-3.1-8b-energy-sft is an 8 billion parameter Llama-3.1-based model, part of the Green Tea replication package. It is specifically fine-tuned for energy-aware code generation using an energy-contrastive supervised fine-tuning approach. This model achieves 5.97% CARET (Correctness-Adjusted Reduction in Energy Total) on a 143-problem benchmark, making it suitable for optimizing code generation for energy efficiency.
Loading preview...
Green Tea Llama-3.1-8B Energy-SFT Overview
This model, saurabh-singh-rajput/green-tea-llama-3.1-8b-energy-sft, is an 8 billion parameter language model built upon the unsloth/Meta-Llama-3.1-8B base architecture. It is a key component of the Green Tea replication package, developed by Rajput and Sharma, focusing on energy-aware code generation.
Key Capabilities & Differentiation
- Energy-Aware Code Generation: Unlike general-purpose code models, this model is specifically trained using an energy-contrastive supervised fine-tuning (SFT) method. This unique training objective prioritizes generating code that is not only correct but also energy-efficient.
- Performance Metric: It achieves a 5.97% CARET (Correctness-Adjusted Reduction in Energy Total) on a dedicated 143-problem held-out benchmark, demonstrating its effectiveness in reducing energy consumption while maintaining code correctness.
- Research-Backed: This model is directly associated with the research paper "Beyond the Need for Speed: Energy-Aware Code Generation via Simulation-Guided Reinforcement Learning".
When to Use This Model
This model is particularly well-suited for use cases where:
- Energy Efficiency is Critical: Developers or researchers are looking to generate code with a focus on minimizing energy consumption, which is vital for edge devices, sustainable computing, or large-scale deployments.
- Code Generation Tasks: The primary application is generating programming code, with an added layer of energy optimization.
- Research in Sustainable AI/Software: It serves as a valuable tool and benchmark for academic and industrial research into energy-efficient software development and AI models.