Easily fine-tune, evaluate and deploy gpt-oss, Qwen3, DeepSeek-R1, or any open source LLM / VLM!

Federated Learning for LLMs.

Upload your data → Get a fine-tuned SLM. Free.

The official implementation of "A Language Modeling Approach to Diacritic-Free Hebrew TTS"

Automatically exported from code.google.com/p/audiotools

Azure IoT Operations with Edge AI in-a-Box

Fine-tuning open-source large and small instruct/chat language models (LLMs & SLMs) from the Hugging Face Model Hub using public datasets.

Kurtis is a fine-tuning, inference and evaluation tool built for SLMs (Small Language Models), such as Huggingface's SmolLM2.

This project demonstrates how to evaluate the use of LLMs and SLMs for extracting structured data from documents using .NET

🧳 SayHalo – AI-powered (gen-ai) SLM Aggregator for seamless chat experience. 💬🤖 Built with Next.js, TypeScript, Tailwind & Resend. ⚡✨ Early access available now! 🔥

This repository documents the journey of building a Small Language Model (SLM) from the ground up using Python and PyTorch. The model is a decoder-only Transformer, trained on a custom corpus of academic papers and articles focused on artificial intelligence theory.

SLM-T-Q-Relay implements a query routing mechanism within a tree of small language models, where each node either answers or forwards queries to the most relevant model.

I'm lazy... That's why I'm doing this...

Deployment of Liquid AI SLM (Small Language Model) as a secure, scalable REST API on AWS using FastAPI, Docker, and Terraform

Este projeto cria um chatbot local em linha de comando, utilizando o modelo quantizado (q4_0) gemma3 via Ollama. Ele oferece uma base sólida para construir interfaces conversacionais sem depender de APIs externas, ideal para aplicações locais, ambientes offline ou situações onde a privacidade dos dados é essencial.

Language model

A lightweight, context-aware language model with function-calling capabilities. Train, predict, and execute in multiple contexts.

This is a lightweight Decoder-Only language model with only 27MB of parameters, optimized for efficient text generation. It features multi-head attention, rotary embeddings, RMSNorm, and KV-cache, supporting fast inference and token generation.