Today's Fine-tuning & Training: Fastest-Growing Projects — May 09, 2026
Today's the Fine-tuning & Training space, we're seeing a surge in interest around innovative methods for optimizing and customizing large language models (LLMs). Developers are eager to explore new techniques for fine-tuning these models to achieve better performance, efficiency, and adaptability. As a result, repositories that offer practical solutions and tools for LLM fine-tuning are gaining significant traction.
raiyanyahya/how-to-train-your-gpt has taken the top spot with an impressive Growth Score of 98.00 and 762 stars. This repository provides a step-by-step guide to building a modern LLM from scratch, with every line commented and explained in an approachable manner. Its massive growth is likely due to its accessibility and comprehensiveness, making it an invaluable resource for developers looking to dive into LLM training.
QingGo/engram-peft boasts an impressive 100 commits over the past 30 days, earning a Growth Score of 12.04 and garnering 32 stars. This repository offers an unofficial implementation of DeepSeek Engram, allowing users to inject high-capacity conditional memory into LLMs via sparse retrieval PEFT without increasing inference FLOPs. Its growth can be attributed to the innovative approach it takes in optimizing LLM performance.
UNfukashigi/Anima-LoRA-Factory has seen significant activity with 45 commits over the past month, resulting in a Growth Score of 7.50 and 27 stars. This user-friendly GUI tool is designed for training LoRAs for next-generation Anima diffusion models, providing an intuitive interface for developers to fine-tune their models. Its growth is likely driven by its ease of use and applicability to cutting-edge AI research.
ZJU-OmniAI/GFT has achieved a Growth Score of 6.65 with 30 stars, thanks in part to its 45 commits over the past month. This repository introduces GFT, a method that fine-tunes LLMs using unbiased group advantages and dynamic coefficient rectification, showcasing its potential for improving model performance. Its growth is likely due to the novelty of its approach and the potential benefits it offers.
semidark/kokoro-deutsch has gained 34 stars and achieved a Growth Score of 4.93, with 36 commits over the past month. This project provides a complete training recipe for fine-tuning Kokoro-82M on German language data, offering a practical solution for developers looking to adapt LLMs to new languages. Its growth is likely driven by its specificity and usefulness in real-world applications.
Jackohhhh/MedLLM-Finetuning boasts 21 stars and a Growth Score of 3.20, with 21 commits over the past month. This repository offers an out-of-the-box fine-tuning framework for medical binary classification tasks using LLMs, providing a valuable resource for developers working in the healthcare sector. Its growth is likely due to its focus on a specific domain and ease of use.
Mintzs/oogaboogalm has seen moderate activity with 12 commits over the past month, resulting in a Growth Score of 2.10 and 47 stars. This project explores fine-tuning LLMs using baked-in "caveman system prompts" to reduce token usage, an innovative approach that's generating interest among developers. Its growth is likely driven by its novelty and potential for improving model efficiency.
hlpun/Train-in-Silence has achieved a Growth Score of 1.31 with 41 stars, thanks in part to its 5 commits over the past month. This repository offers an automated VRAM calculator and Task-Aware MCP server for LLM fine-tuning, helping developers optimize their training processes. Its growth is likely due to its practicality and potential for streamlining development workflows.
raiyanyahya/how-to-train-your-gpt has taken the top spot with an impressive Growth Score of 98.00 and 762 stars. This repository provides a step-by-step guide to building a modern LLM from scratch, with every line commented and explained in an approachable manner. Its massive growth is likely due to its accessibility and comprehensiveness, making it an invaluable resource for developers looking to dive into LLM training.
QingGo/engram-peft boasts an impressive 100 commits over the past 30 days, earning a Growth Score of 12.04 and garnering 32 stars. This repository offers an unofficial implementation of DeepSeek Engram, allowing users to inject high-capacity conditional memory into LLMs via sparse retrieval PEFT without increasing inference FLOPs. Its growth can be attributed to the innovative approach it takes in optimizing LLM performance.
UNfukashigi/Anima-LoRA-Factory has seen significant activity with 45 commits over the past month, resulting in a Growth Score of 7.50 and 27 stars. This user-friendly GUI tool is designed for training LoRAs for next-generation Anima diffusion models, providing an intuitive interface for developers to fine-tune their models. Its growth is likely driven by its ease of use and applicability to cutting-edge AI research.
ZJU-OmniAI/GFT has achieved a Growth Score of 6.65 with 30 stars, thanks in part to its 45 commits over the past month. This repository introduces GFT, a method that fine-tunes LLMs using unbiased group advantages and dynamic coefficient rectification, showcasing its potential for improving model performance. Its growth is likely due to the novelty of its approach and the potential benefits it offers.
semidark/kokoro-deutsch has gained 34 stars and achieved a Growth Score of 4.93, with 36 commits over the past month. This project provides a complete training recipe for fine-tuning Kokoro-82M on German language data, offering a practical solution for developers looking to adapt LLMs to new languages. Its growth is likely driven by its specificity and usefulness in real-world applications.
Jackohhhh/MedLLM-Finetuning boasts 21 stars and a Growth Score of 3.20, with 21 commits over the past month. This repository offers an out-of-the-box fine-tuning framework for medical binary classification tasks using LLMs, providing a valuable resource for developers working in the healthcare sector. Its growth is likely due to its focus on a specific domain and ease of use.
Mintzs/oogaboogalm has seen moderate activity with 12 commits over the past month, resulting in a Growth Score of 2.10 and 47 stars. This project explores fine-tuning LLMs using baked-in "caveman system prompts" to reduce token usage, an innovative approach that's generating interest among developers. Its growth is likely driven by its novelty and potential for improving model efficiency.
hlpun/Train-in-Silence has achieved a Growth Score of 1.31 with 41 stars, thanks in part to its 5 commits over the past month. This repository offers an automated VRAM calculator and Task-Aware MCP server for LLM fine-tuning, helping developers optimize their training processes. Its growth is likely due to its practicality and potential for streamlining development workflows.