# Open-Source AI-Powered Optimization in Scalene: Advancing Python Performance Profiling with DeepSeek-R1 and LLaMA 3.2

arXiv: https://arxiv.org/abs/2502.10299 · HTML: https://mlsanity.com/paper/2502.10299/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 14 Feb 2025.

## Built on by (most cited first, 2 of 2)

- [Optimizing Token Consumption in LLMs: A Nano Surge Approach for Code Reasoning Efficiency](https://mlsanity.com/paper/2504.15989/) (2025)
- [Accelerating Reinforcement Learning Algorithms Convergence using Pre-trained Large Language Models as Tutors With Advice Reusing](https://mlsanity.com/paper/2509.08329/) (2025)

## Abstract (arXiv)

Python's flexibility and ease of use come at the cost of performance inefficiencies, requiring developers to rely on profilers to optimize execution. SCALENE, a high-performance CPU, GPU, and memory profiler, provides fine-grained insights into Python applications while running significantly faster than traditional profilers. Originally, SCALENE integrated OpenAI's API to generate AI-powered optimization suggestions, but its reliance on a proprietary API limited accessibility. This study explores the feasibility of using opensource large language models (LLMs), such as DeepSeek-R1 and Llama 3.2, to generate optimization recommendations within SCALENE. Our evaluation reveals that DeepSeek-R1 provides effective code optimizations comparable to proprietary models. We integrate DeepSeek-R1 into SCALENE to automatically analyze performance bottlenecks and suggest improvements, enhancing SCALENE's utility while maintaining its open-source nature. This study demonstrates that open-source LLMs can be viable alternatives for AI-driven code optimization, paving the way for more accessible and cost-effective performance analysis tools.