Edge Runners 3.2 Hackathon Summary
Event Overview
Edge Runners Hackathon brought together 1,089 participants across 180 teams, all eager to innovate using the cutting-edge capabilities of Meta's Llama 3.2. Over an intense 48-hour challenge, participants developed 45 groundbreaking AI applications, demonstrating the vast potential of Llama 3.2's multimodal processing and real-time AI capabilities.
This hackathon empowered developers to build powerful applications optimized for mobile and IoT devices, leveraging lightweight models designed for edge computing. From text and image processing to augmented reality and advanced image recognition, participants explored a wide range of AI solutions
๐ Hackathon Challenge
The challenge for the Edge Runners Hackathon tasks participants with building an edge compute solution using the full spectrum of Llama models. Developers must create a solution optimized to run on mobile or IoT platforms, utilizing lightweight models like Llama 3.2 1B or 3B for real-time operations. However, teams have two options to enhance their solutions with more powerful models. They can either integrate network-enabled features, allowing their application to tap into larger models, such as Llama 3.1 405B or Llama 3.2โs multimodal models (11B, 90B), for tasks like complex data analysis or multimodal processing. Alternatively, they can use larger models during development for purposes like generating synthetic data or labeling, fine-tuning their smaller models to ensure the final solution is optimized for edge devices. The end goal is to produce a high-performing, edge-ready application that balances local processing with optional network-enhanced AI capabilities.
Side Quest Challenge
The Side Quest Challenge invites participants to develop AI-powered tools for the telecommunications industry, focusing on underserved areas. Teams must use at least two Llama 3.1 or 3.2 models, demonstrating both lightweight and multimodal capabilities. Solutions should creatively address key industry needs like workflow optimization or connectivity improvements, with a focus on real-world impact. Bonus: solutions compatible with edge devices will also qualify for the main track challenge.
๐ค Partners and Technologies
A big shoutout to our partner AI/ML API for providing exclusive access to 500 keys, each with $20 in AI/ML API credits, to help supercharge participants' AI projects during the hackathon! By using the promo code, participants unlocked a free subscription week with access to all Llama 3.2 models, empowering them to build cutting-edge solutions.
With over 100 models, including Llama, Qwen, and Flux, AI/ML API covers a wide range of needs, from image generation to computer vision, all through a single API. This powerful support gave participants the tools they needed to bring their innovative ideas to life. โจ
A huge thanks to AI/ML API for helping our participants elevate their projects with these incredible resources! Be sure to visit their Discord for technical support and maximize the potential of your future AI creations.
๐ Prizes and Opportunities
Edge Runners Hackathon offers exciting rewards for top innovators. The winning team will take home a grand prize of $3,000, while the second-place team earns $1,500, and third place secures $500.
But the prizes donโt stop thereโtop-performing teams will also be fast-tracked for interviews with the prestigious NEXT acceleration program. This 6-week intensive helps startups turn their prototypes into market-ready products. Participants will receive personalized mentorship from industry experts, build a go-to-market strategy, and gain access to essential resources, expanding their professional networks and refining their ideas to prepare for success.
For the Side Quest Challenge, which focuses on AI-powered solutions for the telecommunications industry in underserved areas, participants have the chance to win an invitation, with transport and accommodation covered, to attend our upcoming Giga Hackathons in Barcelona or Geneva in partnership with UNICEF. This is a unique opportunity to showcase your talents on a global stage while helping drive real-world impact.
๐ The Hackathon Winners
๐ฅ1st Place: Text-to-Model Automated Finetuning with Llama-GPTs - is a platform that simplifies fine-tuning LLMs for custom use, turning natural language prompts into datasets to fine-tune models like LLaMA 3.2. It offers offline, private, adaptable AI solutions.
๐ฅ2nd Place: TestFastTrack - enhances IELTS prep with instant, clear feedback on real test content, helping users learn from mistakes. Its visual approach and tracking boost engagement and progress.
๐ฅ3rd Place: GridSense AI - transforms micro-grid management in underserved areas using IoT and edge computing for real-time insights, predictive analytics, and adaptive learning to boost energy efficiency and reliability.
โญ Side Quest Challenge: Thank you to all who participated in the Side Quest Challenge. While we saw innovative approaches, no entries met the full qualification criteria this time. We appreciate your effort and encourage you to continue pushing boundaries in AI development!
๐ฆพ Conclusion
The Edge Runners Hackathon highlighted impressive innovation as teams harnessed Llama models to develop advanced AI solutions for edge devices in underserved telecom areas. Over an intense 48 hours, participants showcased outstanding creativity, resulting in groundbreaking projects. In addition to cash prizes, top teams earned entry to the NEXT acceleration program and potential invites to upcoming Giga Hackathons, with support from partners like AI/ML API, underscoring the boundless potential of AI at the edge.
Thank you to everyone who joined the Side Quest Challenge! Although no entries fully met the criteria, we commend your efforts and encourage continued exploration in AI. Teams will have another opportunity to compete for this prize in future hackathons.
Watch Edge Runners 3.2 Kick-off Stream ๐ https://www.twitch.tv/videos/2279161261
Watch Edge Runners 3.2 Winners Stream ๐ https://www.twitch.tv/videos/2284989614
