Whether you are an amateur, an AI enthusiast, or a veteran ML engineer, we invite you to participate in The Fifth Elephant Open Source AI hackathon. We will have experienced mentors giving you guidance on the projects that you are working on. And did we say cash prizes!?!?!
You can submit your project idea and outline here.
- Only open-source models and projects will be valid. So please use LLaMa 2 model by Meta or other open-source models only.
- Jury members will need a link to a demo, and your GitHub repo for the project.
- A public video link to the demo will also be required. You can use this for the in-person presentation to the jury members as well.
Participants can propose projects which cover the spectrum of GenAI. Following are some of the themes you can work on, and which the jury will consider:
- AI for Scientific Research: e.g. Protein folding models, climate models, drug discovery, image recognition for scientific research, simulations for material science, epidemiology, and more.
- AI for inclusivity and accessibility: e.g. STT/TTS, automated audio descriptions (for non-voice content), automated color blindness correction, AI powered sign language generation, real-time AI powered captioning display for events, educational resources, and content translation across languages by leveraging multi-lingual models, adaptive content for differences in learning ability and/or neurodivergence, etc.
- AI and creative expression: e.g., generative audio, video, text and visuals and ways to combine these in a production-oriented direction, including AR/VR/Gaming and OTT implementations.
- AI in education: e.g., personalized learning plans, adaptive learning plans, content creation, translation with context, AI tutors, productivity tools, well-being improvement tools, etc.
- AI for India: for e.g., India-specific law, models that focus on indic languages, renewable energy optimization, disaster response and relief, education accessibility.
Divya Tak: co-founder at Joyus, a multi-disciplinary creative. Divya also runs the AI for creatives community.
Bharat Shetty is an AI/ML Consultant. He has worked for Airtel Labs and other organizations on AI/ML/NLP platforms and products, across diverse verticals such as conversational AI, EdTech, IOT, and healthcare. Bharat is the editor of The Fifth Elephant Winter edition, and papers discussion community.
All submissions for the projects must be made here. Your ideas and projects are iterative. Use this opportunity to discuss your ideas with the curators and mentors, and improve on them as the hackathon date approaches.
Submissions are also helpful to find collaborators for your projects. Be open and forthcoming in sharing your ideas.
Five prizes of Rs. 1,00,000 (one lakh rupees) each, will be given to winners at the hackathon. One prize is allocated for each theme.
If you have questions about the format of the hackathon, post a comment here.
Follow @fifthel on Twitter.
For any inquiries, call The Fifth Elephant on +91-7676332020.
OpenSource Prompt IDE to create, version, backtest, deploy, share and monetize prompt packages
In LLM world, the prompt is the input to the LLM and the output is generated text/image/audio/video. But when it comes to automation using the prompts, there are many problems faced by developers in Production Environments.
- Accuracy: Prompt on generalised LLMs like OpenAI, llama2 offers lower accuracy from 50-90% depends on what kind of prompts.
- Cost: Increasing the accuracy of prompt takes time, data and gpu, and its iterative and costly affair
- Interoperability: Upgrading the underlying llm or moving to differing variants break the prompt accuracy, so maintainnance become a major headache for enginnering teams
- Reusability: Since there is no way to package and share a prompt, developers around the world are reinventing the wheel.
- Knowledge Upgradation: Base LLMs offers have 6-12 old months information.
Overall 80% of developer bandwith and server cost is spend in fighting LLM Selection, Prompt Fine tunning and achieving production level accuracy.
Sugarcane AI is a Prompt IDE that helps solves all of the problems by bringing the Micro Services architecture in the LLM World. Sugarcane AI offers Open source Prompt IDE to build, version, backtest, deploy, share high performance, and ready to use prompts based packages which are available over APIs. All the prompt executions are logged for further dataset creation and It further enables the train the Task Specific LLMs AKA Micro LLms, based on the prompt Package data. These Micro LLMs helps achieve better performance and accuracy at much lower cost. Similar to npm, a prompt package can be listed on registry and can even be sold on marketplace, enables faster development of LLMs apps in a reliable way.
Sugarcane AI is on a mission to upskill/educate 10 million future prompt developers and which will contribute in the prompt package creationa and maintainance. Using these prompt packages, a LLM can be build upto 20x faster.
With a developer-centric approach, the opensource project focuses on solving prompt accuracy problems and enhancing developer productivity by adopting a microservices and collaborative approach for LLM application development. It will ensure performance and reliability through the open-source community of prompt developers worldwide.
*Prompt Package: An npm-like package containing LLM configs, prompt templates, and underlying datasets and examples.
**Micro LLMs: Task-specific Fine-Tuned LLMs trained on smaller and targeted datasets.
Track: AI in education: With this project, we are offering a Prompt IDE which is easy to use playground for prompt enginners beginners and they can create their first prompt App just by writing a prompt.