Today's Fine-tuning & Training: Fastest-Growing Projects — April 29, 2026
Today's Fine-tuning & Training space saw a surge in innovative approaches to optimizing AI model performance, with many repositories leveraging PyTorch and other cutting-edge technologies. One notable trend is the increasing focus on multimodal fine-tuning, as well as the development of user-friendly tools for training and deploying large language models.
mattmireles/gemma-tuner-multimodal takes the top spot this week with a growth score of 52.18 and 1,393 stars. This repository provides a framework for fine-tuning Gemma 4 and 3n models on Apple Silicon using PyTorch and Metal Performance Shaders, allowing for efficient training on multimodal data such as audio, images, and text. Its impressive growth is likely due to the increasing demand for flexible and high-performance fine-tuning tools.
QingGo/engram-peft comes in second with a growth score of 19.00 and 31 stars. This unofficial implementation of DeepSeek Engram enables users to inject high-capacity conditional memory into large language models via sparse retrieval PEFT, without increasing inference FLOPs. Its growing popularity can be attributed to the ongoing interest in optimizing LLMs for specific tasks while maintaining efficiency.
UNfukashigi/Anima-LoRA-Factory boasts a growth score of 14.20 and 23 stars. This user-friendly GUI tool is designed specifically for training LoRAs for next-generation Anima diffusion models, making it an attractive option for researchers and developers looking to fine-tune their models with ease. Its growth is likely driven by the increasing adoption of diffusion-based architectures in various applications.
ZJU-OmniAI/GFT has a growth score of 11.73 and 29 stars. This repository introduces GFT, a novel approach that combines imitation learning and reward fine-tuning using unbiased group advantages and dynamic coefficient rectification. Its growing popularity may be attributed to the ongoing quest for more effective and efficient fine-tuning methods.
SUM-INNOVATION/RUMUS has a growth score of 9.41 and an impressive 201 stars. This Rust-based framework provides a flexible and modular way to train neural networks, making it an attractive option for developers seeking a lightweight and efficient solution. Its growth is likely driven by the increasing adoption of Rust in AI development.
semidark/kokoro-deutsch boasts a growth score of 6.39 and 30 stars. This project offers a complete training recipe for fine-tuning Kokoro-82M on German, providing a valuable resource for researchers looking to adapt this model to new languages. Its growth is likely driven by the ongoing interest in multilingual AI models.
Dynamis-Labs/spectralquant has a growth score of 4.04 and 128 stars. This repository introduces a novel approach that breaks TurboQuant's compression limit via spectral structure, offering insights into efficient quantization techniques for neural networks. Its growth is likely driven by the ongoing quest for more efficient model deployment strategies.
Mintzs/oogaboogalm has a growth score of 3.16 and 45 stars. This project explores the idea of fine-tuning AI models to reduce token usage, offering an innovative approach to optimizing language models. Its growing popularity may be attributed to the increasing focus on efficiency in NLP applications.
PentesterFlow/OffensiveSET boasts a growth score of 3.02 and 73 stars. This repository provides a dataset generator for high-quality pentesting conversation datasets, allowing users to fine-tune their LLMs for specific security-related tasks. Its growth is likely driven by the increasing demand for AI-powered security solutions.
Goekdeniz-Guelmez/moshi-finetune-mlx has a growth score of 0.71 and 25 stars. This repository offers a tool for fine-tuning Moshi models on Apple Silicon, providing an efficient solution for speech-to-speech applications. Its growth may be driven by the increasing adoption of real-time speech processing in various industries.
mattmireles/gemma-tuner-multimodal takes the top spot this week with a growth score of 52.18 and 1,393 stars. This repository provides a framework for fine-tuning Gemma 4 and 3n models on Apple Silicon using PyTorch and Metal Performance Shaders, allowing for efficient training on multimodal data such as audio, images, and text. Its impressive growth is likely due to the increasing demand for flexible and high-performance fine-tuning tools.
QingGo/engram-peft comes in second with a growth score of 19.00 and 31 stars. This unofficial implementation of DeepSeek Engram enables users to inject high-capacity conditional memory into large language models via sparse retrieval PEFT, without increasing inference FLOPs. Its growing popularity can be attributed to the ongoing interest in optimizing LLMs for specific tasks while maintaining efficiency.
UNfukashigi/Anima-LoRA-Factory boasts a growth score of 14.20 and 23 stars. This user-friendly GUI tool is designed specifically for training LoRAs for next-generation Anima diffusion models, making it an attractive option for researchers and developers looking to fine-tune their models with ease. Its growth is likely driven by the increasing adoption of diffusion-based architectures in various applications.
ZJU-OmniAI/GFT has a growth score of 11.73 and 29 stars. This repository introduces GFT, a novel approach that combines imitation learning and reward fine-tuning using unbiased group advantages and dynamic coefficient rectification. Its growing popularity may be attributed to the ongoing quest for more effective and efficient fine-tuning methods.
SUM-INNOVATION/RUMUS has a growth score of 9.41 and an impressive 201 stars. This Rust-based framework provides a flexible and modular way to train neural networks, making it an attractive option for developers seeking a lightweight and efficient solution. Its growth is likely driven by the increasing adoption of Rust in AI development.
semidark/kokoro-deutsch boasts a growth score of 6.39 and 30 stars. This project offers a complete training recipe for fine-tuning Kokoro-82M on German, providing a valuable resource for researchers looking to adapt this model to new languages. Its growth is likely driven by the ongoing interest in multilingual AI models.
Dynamis-Labs/spectralquant has a growth score of 4.04 and 128 stars. This repository introduces a novel approach that breaks TurboQuant's compression limit via spectral structure, offering insights into efficient quantization techniques for neural networks. Its growth is likely driven by the ongoing quest for more efficient model deployment strategies.
Mintzs/oogaboogalm has a growth score of 3.16 and 45 stars. This project explores the idea of fine-tuning AI models to reduce token usage, offering an innovative approach to optimizing language models. Its growing popularity may be attributed to the increasing focus on efficiency in NLP applications.
PentesterFlow/OffensiveSET boasts a growth score of 3.02 and 73 stars. This repository provides a dataset generator for high-quality pentesting conversation datasets, allowing users to fine-tune their LLMs for specific security-related tasks. Its growth is likely driven by the increasing demand for AI-powered security solutions.
Goekdeniz-Guelmez/moshi-finetune-mlx has a growth score of 0.71 and 25 stars. This repository offers a tool for fine-tuning Moshi models on Apple Silicon, providing an efficient solution for speech-to-speech applications. Its growth may be driven by the increasing adoption of real-time speech processing in various industries.