Today's Fine-tuning & Training: Fastest-Growing Projects — May 05, 2026
Today's Fine-tuning & Training, we've seen a surge of interest in tools that enable efficient and effective training of large language models (LLMs). Many developers are turning to fine-tuning techniques to adapt pre-trained models to specific tasks or domains, while others are exploring new architectures and methods for improving model performance. As a result, repositories focused on fine-tuning and training have seen significant growth in popularity.
Matts Mireles' gemma-tuner-multimodal repository has taken the top spot with an impressive Growth Score of 41.30 and over 1,407 stars. This tool allows users to fine-tune Gemma 4 and 3n models on Apple Silicon using PyTorch and Metal Performance Shaders, making it an attractive option for those looking to leverage multimodal inputs (audio, images, and text). Its high growth rate is likely due to its unique combination of hardware and software optimization.
QingGo's engram-peft repository has also seen significant traction with a Growth Score of 14.04 and 31 stars. This unofficial implementation of DeepSeek Engram enables the injection of high-capacity conditional memory into LLMs via sparse retrieval PEFT, without increasing inference FLOPs. Its growth is likely driven by interest in improving model performance while maintaining efficiency.
UNfukashigi's Anima-LoRA-Factory has gained attention with a Growth Score of 9.31 and 25 stars. This user-friendly GUI tool is designed for training LoRAs (Low-Rank Adaptation) for next-generation Anima diffusion models, making it an appealing option for those interested in generative AI. Its growth rate suggests a strong demand for accessible tools that simplify complex model training tasks.
ZJU-OmniAI's GFT repository has achieved a Growth Score of 8.05 and 30 stars. This tool focuses on fine-tuning with unbiased group advantages and dynamic coefficient rectification, offering an innovative approach to improving model performance. Its growth is likely driven by interest in exploring new techniques for efficient training.
Semidark's kokoro-deutsch repository has gained popularity with a Growth Score of 5.26 and 32 stars. This project provides a complete recipe for fine-tuning Kokoro-82M on the German language, making it an attractive option for those interested in multilingual model adaptation. Its growth rate suggests a strong demand for resources that facilitate language-specific training.
Raiyanyahya's how-to-train-your-gpt repository has maintained its popularity with a Growth Score of 4.69 and over 338 stars. This comprehensive guide to building an LLM from scratch provides every line commented and explained in simple terms, making it an excellent resource for beginners and experienced developers alike. Its enduring growth is likely due to the value it offers as a educational tool.
The remaining repositories, including Jackohhhh's MedLLM-Finetuning, Mintzs' oogaboogalm, and hlpun's Train-in-Silence, have also seen notable growth, albeit at a slower pace. These tools offer specialized solutions for tasks such as medical binary classification, token-efficient model design, and automated VRAM calculation, highlighting the diversity of interests within the Fine-tuning & Training space.
Matts Mireles' gemma-tuner-multimodal repository has taken the top spot with an impressive Growth Score of 41.30 and over 1,407 stars. This tool allows users to fine-tune Gemma 4 and 3n models on Apple Silicon using PyTorch and Metal Performance Shaders, making it an attractive option for those looking to leverage multimodal inputs (audio, images, and text). Its high growth rate is likely due to its unique combination of hardware and software optimization.
QingGo's engram-peft repository has also seen significant traction with a Growth Score of 14.04 and 31 stars. This unofficial implementation of DeepSeek Engram enables the injection of high-capacity conditional memory into LLMs via sparse retrieval PEFT, without increasing inference FLOPs. Its growth is likely driven by interest in improving model performance while maintaining efficiency.
UNfukashigi's Anima-LoRA-Factory has gained attention with a Growth Score of 9.31 and 25 stars. This user-friendly GUI tool is designed for training LoRAs (Low-Rank Adaptation) for next-generation Anima diffusion models, making it an appealing option for those interested in generative AI. Its growth rate suggests a strong demand for accessible tools that simplify complex model training tasks.
ZJU-OmniAI's GFT repository has achieved a Growth Score of 8.05 and 30 stars. This tool focuses on fine-tuning with unbiased group advantages and dynamic coefficient rectification, offering an innovative approach to improving model performance. Its growth is likely driven by interest in exploring new techniques for efficient training.
Semidark's kokoro-deutsch repository has gained popularity with a Growth Score of 5.26 and 32 stars. This project provides a complete recipe for fine-tuning Kokoro-82M on the German language, making it an attractive option for those interested in multilingual model adaptation. Its growth rate suggests a strong demand for resources that facilitate language-specific training.
Raiyanyahya's how-to-train-your-gpt repository has maintained its popularity with a Growth Score of 4.69 and over 338 stars. This comprehensive guide to building an LLM from scratch provides every line commented and explained in simple terms, making it an excellent resource for beginners and experienced developers alike. Its enduring growth is likely due to the value it offers as a educational tool.
The remaining repositories, including Jackohhhh's MedLLM-Finetuning, Mintzs' oogaboogalm, and hlpun's Train-in-Silence, have also seen notable growth, albeit at a slower pace. These tools offer specialized solutions for tasks such as medical binary classification, token-efficient model design, and automated VRAM calculation, highlighting the diversity of interests within the Fine-tuning & Training space.