Today's Fine-tuning & Training: Fastest-Growing Projects — April 24, 2026
Today's the Fine-tuning & Training space, we're seeing a surge of innovative projects leveraging cutting-edge techniques to optimize and fine-tune various AI models. From multimodal training to sparse retrieval PEFT, developers are pushing the boundaries of what's possible with large language models (LLMs). With many repositories showcasing significant growth, it's clear that the community is eager to explore new approaches to fine-tuning and training.
mattmireles/gemma-tuner-multimodal has taken the top spot this week, boasting a Growth Score of 66.88 and an impressive 1,380 stars. This project enables users to fine-tune Gemma 4 and 3n with audio, images, and text on Apple Silicon using PyTorch and Metal Performance Shaders, making it an attractive solution for those looking to explore multimodal training. Its rapid growth can be attributed to the increasing interest in multimodal learning and the project's well-documented approach.
QingGo/engram-peft has also seen significant growth, with a Growth Score of 26.67 and 31 stars. This unofficial implementation of DeepSeek Engram injects high-capacity conditional memory into LLMs via sparse retrieval PEFT without increasing inference FLOPs. Its growing popularity can be attributed to the innovative approach it takes to addressing the limitations of traditional fine-tuning methods.
ZJU-OmniAI/GFT is another notable project, with a Growth Score of 18.75 and 27 stars. This repository presents a novel approach to fine-tuning called GFT, which leverages unbiased group advantages and dynamic coefficient rectification. Its growth can be attributed to the interest in exploring new methods for improving the efficiency and effectiveness of fine-tuning.
WillowHe/EvoOpt_oppangu_optimization_model has garnered significant attention with 514 stars, despite a relatively low Growth Score of 10.70. This project provides solutions leveraging Openpangu - 7B as the base model for fine-tuning and application of LLMs in operations research optimization tasks. Its popularity can be attributed to the growing interest in applying AI to real-world optimization problems.
SUM-INNOVATION/RUMUS, with a Growth Score of 9.46 and 139 stars, offers a Rust-based framework for training neural networks. Its growth can be attributed to the increasing adoption of Rust as a programming language for building high-performance applications.
semidark/kokoro-deutsch has seen moderate growth with a Growth Score of 8.39 and 28 stars. This project provides a complete, documented training recipe for fine-tuning Kokoro-82M on German. Its growth can be attributed to the interest in exploring new languages and models for natural language processing tasks.
The remaining projects, while showing promise, have seen slower growth this week. Dynamis-Labs/spectralquant (Growth Score: 4.89, Stars: 123) presents an innovative approach to breaking compression limits via spectral structure, but its growth has been relatively slow. Mintzs/oogaboogalm (Growth Score: 4.25, Stars: 44) explores fine-tuning AI models to reduce token use, but its growth has been steady rather than rapid.
Overall, Today's Fine-tuning & Training space is characterized by innovative approaches to optimizing and fine-tuning various AI models. As the community continues to explore new techniques and methods, we can expect to see further growth and development in this exciting area of research.
mattmireles/gemma-tuner-multimodal has taken the top spot this week, boasting a Growth Score of 66.88 and an impressive 1,380 stars. This project enables users to fine-tune Gemma 4 and 3n with audio, images, and text on Apple Silicon using PyTorch and Metal Performance Shaders, making it an attractive solution for those looking to explore multimodal training. Its rapid growth can be attributed to the increasing interest in multimodal learning and the project's well-documented approach.
QingGo/engram-peft has also seen significant growth, with a Growth Score of 26.67 and 31 stars. This unofficial implementation of DeepSeek Engram injects high-capacity conditional memory into LLMs via sparse retrieval PEFT without increasing inference FLOPs. Its growing popularity can be attributed to the innovative approach it takes to addressing the limitations of traditional fine-tuning methods.
ZJU-OmniAI/GFT is another notable project, with a Growth Score of 18.75 and 27 stars. This repository presents a novel approach to fine-tuning called GFT, which leverages unbiased group advantages and dynamic coefficient rectification. Its growth can be attributed to the interest in exploring new methods for improving the efficiency and effectiveness of fine-tuning.
WillowHe/EvoOpt_oppangu_optimization_model has garnered significant attention with 514 stars, despite a relatively low Growth Score of 10.70. This project provides solutions leveraging Openpangu - 7B as the base model for fine-tuning and application of LLMs in operations research optimization tasks. Its popularity can be attributed to the growing interest in applying AI to real-world optimization problems.
SUM-INNOVATION/RUMUS, with a Growth Score of 9.46 and 139 stars, offers a Rust-based framework for training neural networks. Its growth can be attributed to the increasing adoption of Rust as a programming language for building high-performance applications.
semidark/kokoro-deutsch has seen moderate growth with a Growth Score of 8.39 and 28 stars. This project provides a complete, documented training recipe for fine-tuning Kokoro-82M on German. Its growth can be attributed to the interest in exploring new languages and models for natural language processing tasks.
The remaining projects, while showing promise, have seen slower growth this week. Dynamis-Labs/spectralquant (Growth Score: 4.89, Stars: 123) presents an innovative approach to breaking compression limits via spectral structure, but its growth has been relatively slow. Mintzs/oogaboogalm (Growth Score: 4.25, Stars: 44) explores fine-tuning AI models to reduce token use, but its growth has been steady rather than rapid.
Overall, Today's Fine-tuning & Training space is characterized by innovative approaches to optimizing and fine-tuning various AI models. As the community continues to explore new techniques and methods, we can expect to see further growth and development in this exciting area of research.