LLM-PowerHouse: Unleash LLMs' potential through curated tutorials, best practices, and ready-to-use code for custom training and inferencing.

Autonomously train research-agent LLMs on custom data using reinforcement learning and self-verification.

This is a LangChain LLM template that allows you to train your own custom AI LLM.

A dataset for training customer service chatbot models on LLMs

A complete end-to-end pipeline for training specialized Small Language Models (SLMs) on custom business data. OTTO enables organizations to train domain-specific language models without relying on expensive LLM fine-tuning or external APIs.

No description available

Automated static analysis tool powered by custom code-trained LLMs.

Implementation of ICLR 2025 paper "Q-Adapter: Customizing Pre-trained LLMs to New Preferences with Forgetting Mitigation"

A 4-hour long tutorial session for learning to use LLMs and align them with custom data. We will also train a custom LLM.

A tiny language model (LLM) written in pure Dart. It implements a minimal GPT-style network with a custom Tensor/autograd engine, trains on a small built‑in text sample, and prints a sampled continuation.

This repository features a custom-built decoder-only language model (LLM) with a total of 37 million parameters 🔥. I train the model to be able to ask question from a given context

Custom Trained LLM application with Llama, and grounding via RAG. This project uses Streamlit to create a simple UX LLM based chatbot with Llama3 & RAG grounding on Stehen Hawking's books

An AI-powered Telegram chatbot that engages users in flirty conversation and drives them to your content page (Fanvue, OnlyFans, or any link). Uses a custom DPO-trained LLM (via RunPod Serverless) for natural chat, with intent detection, lead scoring, photo escalation, and smart CTA timing.

Groq QA is a CLI tool and Python library for generating question-answer pairs from text, featuring custom train/test dataset split ratios for fine-tuning large language models (LLMs).

Built an AI-powered chatbot using Python, RAG, and the Gemini API to assist over 500 users in making safer decisions, achieving 90%+ response accuracy and improving public safety. Additionally, developed a legal assistant LLM trained on custom data using RAG and ChromaDB, offering 85%+ accurate legal guidance to more than 300 users.

Creating a LLM based Model which can be trained with a given custom PDF and answer questions using it

Fine-tuned a Large Language Model (LLM) on domain-specific data using Hugging Face Transformers and Trainer API. Implemented training optimizations with LoRA and evaluated model performance using custom prompts. Deployed the model using Streamlit with a visually appealing UI and hosted it on Hugging Face Spaces for real-time interaction.

Custom trained LLM for SpacedAce's e-Learning platform

Easiest way to create, train, and customize SoTA LLMs

GPT 3.5 based LLM trained on custom data based on Age of Empires 2 DE posts, videos and messages from various sources

AI web application, using custom data-set to train and build a LLM from scratch to provide assistance for fraud detection and protfolio management

A program that enhances and customizes ChatGPT's underlying pre-trained LLM w/ transformer architecture. Based on OpenAI's beta InstructGPT fine-tune model.

This is the official UI for UrduGPT — a custom-built English → Urdu translator powered by a Transformer-based LLM trained from scratch using PyTorch.

Build and train a custom GPT model from scratch using PyTorch. Includes BPE tokenizer, transformer architecture, and text generation. Ideal for learning LLMs, NLP, and training generative AI without pre-trained models. Great for research, education, and hands-on exploration.

This library offers tools to easily fine-tune large language models (LLMs) on custom datasets. It enhances pre-trained models for specific tasks like text generation, classification, translation, and summarization. With a focus on flexibility and simplicity, it enables efficient, specialized model training for various applications.

AI-powered air-to-surface EO/IR classification system using RGB-IR fusion for military object detection. Features YOLOv8 trained on 1500+ custom-annotated IR images of tanks, vehicles, and ships. Integrates with War Thunder via MCP server for real-time simulation control and LLM-assisted target acquisition

Spamlyser Pro is a powerful, real-time SMS spam detection platform built with Streamlit and Transformers, backed by 4 custom-trained LLM backbones DistilBERT, BERT, RoBERTa, and ALBERT. It allows users to classify SMS messages as SPAM or HAM, visualise spam distribution, and analyse risky message features with an elegant, responsive UI.

Train LLM with custom data

This repo contains the code to train and fine-tune the custom LLM with domain specific data. It also contains a few examples to build the chatbot using open-source LLM.

Custom trained LLM Langgraph Chatbot on finance and text based dataset