Today's Fine-tuning & Training: Fastest-Growing Projects — April 28, 2026
Today's Fine-tuning & Training, we're seeing a surge of interest in multimodal fine-tuning and low-resource training methods. Projects that enable efficient training on diverse data types, such as images, audio, and text, are gaining traction among developers. Meanwhile, innovations in parameter-efficient fine-tuning (PEFT) continue to attract attention.
mattmireles/gemma-tuner-multimodal takes the top spot with a growth score of 54.64 and 1,392 stars. This repository provides a framework for fine-tuning Gemma 4 and 3n models on Apple Silicon using PyTorch and Metal Performance Shaders, allowing developers to efficiently train multimodal models. Its rapid growth is likely due to the increasing demand for AI applications that can handle diverse data types.
QingGo/engram-peft has a growth score of 20.19 and 31 stars. This unofficial implementation of DeepSeek Engram injects high-capacity conditional memory into large language models (LLMs) via sparse retrieval PEFT, without increasing inference FLOPs. Its growing popularity may be attributed to the interest in efficient fine-tuning methods that can enhance LLM performance without significant computational overhead.
UNfukashigi/Anima-LoRA-Factory boasts a growth score of 15.78 and 23 stars. This user-friendly GUI tool is designed for training LoRAs (Low-Rank Adaptation) for next-generation Anima diffusion models, making it easier for developers to adapt these models to specific tasks. Its growth may be driven by the increasing adoption of diffusion models in various applications.
ZJU-OmniAI/GFT has a growth score of 12.71 and 29 stars. This repository presents GFT, a method that combines imitation learning with reward fine-tuning using unbiased group advantages and dynamic coefficient rectification. Its growing interest may be due to the need for more efficient and effective training methods in reinforcement learning.
SUM-INNOVATION/RUMUS has a growth score of 9.64 and 195 stars. This Rust-based framework provides a simple way to train neural networks, making it an attractive option for developers looking for alternative deep learning frameworks. Its growth may be attributed to the increasing popularity of Rust as a programming language.
WillowHe/EvoOpt_oppangu_optimization_model has a growth score of 9.22 and 514 stars. This repository provides solutions for fine-tuning Openpangu-7B, a large language model, on operations research optimization tasks. Its growing interest may be driven by the need for more efficient optimization methods in various industries.
semidark/kokoro-deutsch has a growth score of 6.75 and 30 stars. This project provides a complete training recipe for fine-tuning Kokoro-82M on German, showcasing its potential for adapting pre-trained models to new languages. Its growth may be due to the increasing demand for multilingual AI applications.
Dynamis-Labs/spectralquant has a growth score of 4.22 and 128 stars. This repository presents a method that breaks TurboQuant's compression limit via spectral structure, allowing for more efficient model quantization. Its growing interest may be attributed to the need for more efficient model deployment methods.
Mintzs/oogaboogalm has a growth score of 3.33 and 45 stars. This project explores fine-tuning AI models with baked-in skills and prompts to reduce token use, showcasing its potential for creating more efficient language models. Its growing interest may be driven by the need for more efficient natural language processing methods.
PentesterFlow/OffensiveSET has a growth score of 3.13 and 73 stars. This repository provides an Offensive Security Dataset Generator for generating high-quality pentesting conversation datasets for LLM fine-tuning, making it an attractive option for security researchers. Its growing interest may be attributed to the increasing demand for AI-powered security tools.
mattmireles/gemma-tuner-multimodal takes the top spot with a growth score of 54.64 and 1,392 stars. This repository provides a framework for fine-tuning Gemma 4 and 3n models on Apple Silicon using PyTorch and Metal Performance Shaders, allowing developers to efficiently train multimodal models. Its rapid growth is likely due to the increasing demand for AI applications that can handle diverse data types.
QingGo/engram-peft has a growth score of 20.19 and 31 stars. This unofficial implementation of DeepSeek Engram injects high-capacity conditional memory into large language models (LLMs) via sparse retrieval PEFT, without increasing inference FLOPs. Its growing popularity may be attributed to the interest in efficient fine-tuning methods that can enhance LLM performance without significant computational overhead.
UNfukashigi/Anima-LoRA-Factory boasts a growth score of 15.78 and 23 stars. This user-friendly GUI tool is designed for training LoRAs (Low-Rank Adaptation) for next-generation Anima diffusion models, making it easier for developers to adapt these models to specific tasks. Its growth may be driven by the increasing adoption of diffusion models in various applications.
ZJU-OmniAI/GFT has a growth score of 12.71 and 29 stars. This repository presents GFT, a method that combines imitation learning with reward fine-tuning using unbiased group advantages and dynamic coefficient rectification. Its growing interest may be due to the need for more efficient and effective training methods in reinforcement learning.
SUM-INNOVATION/RUMUS has a growth score of 9.64 and 195 stars. This Rust-based framework provides a simple way to train neural networks, making it an attractive option for developers looking for alternative deep learning frameworks. Its growth may be attributed to the increasing popularity of Rust as a programming language.
WillowHe/EvoOpt_oppangu_optimization_model has a growth score of 9.22 and 514 stars. This repository provides solutions for fine-tuning Openpangu-7B, a large language model, on operations research optimization tasks. Its growing interest may be driven by the need for more efficient optimization methods in various industries.
semidark/kokoro-deutsch has a growth score of 6.75 and 30 stars. This project provides a complete training recipe for fine-tuning Kokoro-82M on German, showcasing its potential for adapting pre-trained models to new languages. Its growth may be due to the increasing demand for multilingual AI applications.
Dynamis-Labs/spectralquant has a growth score of 4.22 and 128 stars. This repository presents a method that breaks TurboQuant's compression limit via spectral structure, allowing for more efficient model quantization. Its growing interest may be attributed to the need for more efficient model deployment methods.
Mintzs/oogaboogalm has a growth score of 3.33 and 45 stars. This project explores fine-tuning AI models with baked-in skills and prompts to reduce token use, showcasing its potential for creating more efficient language models. Its growing interest may be driven by the need for more efficient natural language processing methods.
PentesterFlow/OffensiveSET has a growth score of 3.13 and 73 stars. This repository provides an Offensive Security Dataset Generator for generating high-quality pentesting conversation datasets for LLM fine-tuning, making it an attractive option for security researchers. Its growing interest may be attributed to the increasing demand for AI-powered security tools.