🔥 48 hours of creation
🦾 Build with Open Source AI models
🚀 Tutorials and mentors to support you
🤝 Collaborate with the community of AI creators!


Hackathon Winner

Finalists

Olesia Zinchenko

NewNative

Nasti

SimonOlsn

CzechPawel

fabi.s

LizaM

Hai The Dude

Open Source AI Hackathon

Team of ML engineers excited to hack together interesting projects, make new friends, and learn!

Samosa Squad

Karabo

lijnati

Team of devs ML and web devs who love dosa and chutney!

Code Chutney

pi227

rtindru

saniya

Meghraj Bendre

Team LangPain aims to develop innovative AI solutions that address real-world problems

Team LangPain

Abhishek Ranjan

root_admin

Beetho07

Authority101

dayveedben

Oswaldshoo

The best way to predict the future, is to build it.

FINCA

Dgflo

Large language models (LLMs) have transformed the way we approach machine learning and artificial intelligence. These models are incredibly powerful, but their true potential lies in their ability to work in harmony with other sources of data and computation. This library has been developed to assist developers in building applications that combine LLMs with other technologies, allowing them to create truly intelligent and innovative products.

Langchain

Meta is a US technology company that owns the social networks Facebook and Instagram, the instant messaging apps WhatsApp and Messenger, and the virtual reality device brand Meta Quest.

Stability AI

Welcome on Open Source AI Hackathon by lablab.ai! Let's create and build together over next 48 hours!



Event Kick-Off

Let's meet our founders and the team behind the scenes 🤩

Lovely lablab.ai intro

We will talk about the Hackathon challenge, all the technologies and not only for next 48 hours of innovating and coming up with even crazy ideas!

Introduction to the challenge

Let's guide through the platform and Discord server! It's the best time to get all the answers for questions you may have

Hackathon Guide

The live coding session with Hai Nghiem, developer, LangChain builder and lablab.ai community member!

Code Workshop: Make a Spotify AI Agent

Make sure you follow the deadline in CET, and don't forget about mentors help, if needed 🚀

Submission Deadline!

Build your product with the best Open Source AI models!

CoPilot-J is an open-source GitHub Copilot-X an alternative for VSCode. It can chat with code, generates, explains, and refactors code using LocalAI, and is compatible with any open-source model (ggml compatible).

for the implementation We have 2 main components, the first one is VS code plugin that does all the interaction, it is able to chat with models, send code to the conversation, or detect code snippets from the conversation.
the second part is go-skynet/LocalAI which does all the magic of serving AI models, in this demo, we’re using wizard-lm inference to demo the plugin, but you can use any ggml model to use it.

the repo is up here: https://github.com/badgooooor/localai-vscode-plugin.git

PS. Initially we thought we'd use GPT4ALL-J hence the name Copilot-J 😂

CoPilot-J is an open-source GitHub Copilot-X an alternative for VSCode. It can chat with code, generates, explains, and refactors code using LocalAI, and is compatible with any open-source model (ggml compatible).

Copilot-J

Nonthakon Jitchiranant

Yuttakhan B.

Pitch Analyzer allows you to receive a complete pitch from a startup in text format, analyzes the content, and then simulates a conversation between a potential investor and the founder of the startup, answering all questions based on the delivered pitch, or inferring an answer if You can't find the appropriate information. All this is with the intention of preparing the founders for specific questions that they may receive at this type of event if they wish to attract and receive funds for their projects.
Currently, training to make good pitches for founders is complicated by time, sometimes at a cost that startups in earlier stages cannot, and sometimes not adapted to the types of startups that participate, therefore, Pitch Analyzer is the ideal solution for it.

Startup pitch analysis and generation of questions between an investor and a founder.

Pitch Analyzer

Néstor Nicolás Campos Rojas

CAMEL stands for Communicative Agents for "Mind" Exploration of Large Scale Language Model Society. The purpose of this framework is to enhance the collaboration among AI chat agents to achieve tasks with minimal human involvement.

CAMEL

OpenAI is an AI research and deployment company

OpenAI

LangChain

Have you ever been wronged by someone, and you wanted to yell and scream at them, but you held back? You know it's not fair, but you don't want to risk the consequences.
Enter RageGPT! It's an AI that lets you yell and scream at an AI of your adversary, without any consequences. You can type in all caps, swear, and say anything you want. No one will know!
Or use it to practice negotiations for a pay raise or promotion.  Or practice debating your boss / teacher / parent.  Anything!
Want to confront a former lover but not sure what to say?  Try it out at RageGPT and see what happens?  Great no-consequence practice for important life-changing events.

RageGPT: Let Your Anger Out on an AI of Your Adversary! With no consequences!

Mr Barnard Room 12

sjdthree

ChatGPT is a large language model trained by OpenAI to generate human-like text in a conversational style. It is a variant of the GPT-3 model, which was specifically designed to be used to generate text in response to user input. The ChatGPT model has been trained on a vast amount of text data, including conversations and other types of human-generated text, which allows it to generate text that is similar in style and content to human conversation. ChatGPT can be used to generate responses to questions, code, make suggestions, or provide information in a conversational manner, and it is able to do so in a way that is often indistinguishable from human-generated text. The initial model has been trained using Reinforcement Learning from Human Feedback (RLHF), using methods similar to InstructGPT, but with slight differences in the data collection setup. The model is trained using supervised fine-tuning, where human AI trainers provided conversations in which they played both sides—the user and an AI assistant. The trainers would have had access to model-written suggestions to help them compose their responses.

ChatGPT

RageGPT

Users often struggle to find the most relevant and useful documents related to their needs due to Inefficient keyword-based search and Lack of query context. Our solution to this problem is building a tool call Doc Search AI. It is an AI Document Searching tool that addresses the shortcomings of traditional search methods and offers users a more effective and efficient way to find relevant documents by leveraging advanced AI techniques. Doc Search AI will takes users query as input, then using large language model (we use GPT2 open source model) to expand the users' document queries in more details, and use other AI techniques like keywords extraction and vectorize the search query to improve the quality of search so that the results return to users will be more relevant to what users want.


Using AI to Revolutionizes Documents Retrieval for Highly Relevant and Accurate Results.


kopi c peng

Wayne_Myo

LeonTan

Khant

GPT-3 stands for Generative Pre-trained Transformer 3 and it is an autoregressive language model that uses deep learning to produce human-like text. It is the third-generation language prediction model in the GPT-n series created by OpenAI. GPT-3 is currently in open beta.

Open Source AI Hackathon Summary

Hackathon Overview

Winners and Finalists

Copilot-J

Speakers, Mentors, and Organizers

Prizes

Hackathon Winner

Finalists

Submitted Concepts, Prototypes and Pitches

Copilot-J

Pitch Analyzer

RageGPT

DocSearch AI