🌎 Overview

The Lilypad Network is a serverless, distributed compute network providing internet-scale data processing for AI and other next-gen use cases.

At the core of Lilypad's architecture is a combination of off-chain computation and on-chain verification. Participants can buy and sell computing services through a distributed network of GPU nodes, with blockchain technology ensuring trust and transparency. This innovative approach unlocks new revenue streams for hardware providers while optimizing resource utilization across the network.

The Lilypad Network Incentivized Testnet, aka “IncentiveNet”, continues to provide an environment to stress test network economics and ensure our architecture can handle the growing scale of GPU supply. IncentiveNet has reached over 1100 nodes across the globe to date! The Lilypad Network Resource Provider (RP) Leaderboard provides an overview of RP capacity on the network as well as points earned by each address.

During IncentiveNet, the Lilypad team has put a focus on partnering with fantastic organizations and individuals for distribution of Lilypad compute services. From teams like Rare Compute to BioML researchers, we are thrilled to see increasing demand for the compute capacity on IncentiveNet!

The IncentiveNet represents a significant milestone in establishing the critical compute infrastructure necessary for the Lilypad Mainnet to flourish upon launch. By participating in IncentiveNet, users can contribute to the development of the Lilypad ecosystem and earn Lilybit_ rewards, redeemable for the Lilypad ERC20 token upon Mainnet launch.

Sign up to the Lilypad Newsletter and follow us on X!

⚒️ Engineering Update

Our core, web3, and infrastructure teams have been focused on quickly delivering updates for the IncentiveNet while implementing learnings into the economics and architecture of the Lilypad whitepaper and Mainnet roadmap.

As the network GPU supply grows, our team has worked tirelessly to ensure the increasing scale can be handled! We have implemented Telemetry to monitor jobs running on the network and have created a process internally to investigate issues that arise. Community facing tooling will soon be released to provide transparency into Lilybit rewards earned by RPs each day.

Stay connected on Discord and follow along on X to get the latest development announcements from the Lilypad Network!

RPs update to v2.5.0 - Aristotle update

Excellence is not an art. It is the habit of practice.- Aristotle

A network update, Aristotle, was released requiring Bacalhau v1.3.2 be installed and started before the Lilypad-resource-provider (RP) can be started. The RP will only start if Bacalhau is running and at the correct version. Lilypad Resource Providers will need to update to the latest release, Lilypad v2.5.0 Aristotle.

Our team has noticed some RPs in the network are not running Bacalhau (active) or are running the wrong Bacalhau version. This will cause job runs to fail on the respective RP. This update will ensure that the entire network has the correct version of Bacalhau installed and running and ensure reliability for users.

Targeting a node to run a job

In the latest Lilypad v2.5.0 Aristotle update, a new CLI flag --target was added that allows a job creator to send a job to a specific RP for processing. The job creator must simply know the ETH wallet address for the required RP.

This functionality has been included for a variety of stakeholders. Internally, the ability to send specific RPs a job helps our team with network testing, improved module development, and more! Partners running on the IncentiveNet have also requested this functionality for job testing and general operational needs.

🔬 Research Updates

The research team continues to support the upcoming Lilypad Whitepaper, working with a variety of Lilypad partners, and testing modules with the Machine Learning (ML) research community. This critical work provides our team valuable insights into users' experiences building on the Lilypad network and short feedback loops with the Lilypad community.

Lilypad is building for end users in need of accessible, high performance GPU compute. Working with the ML research community and partners like Rare Compute from an early stage in our development has been critical to defining the way the network and associated products need to scale.

RFDiffusion on Lilypad - “Midjourney for Molecules”

Our team has been working with the Bio ML research community to provide tooling that helps solve problems focused on advancing medical research! In the latest collaboration, a Lilypad module has been created to run RFDiffusion and is currently in the test stages of running on the network. RFDiffusion is a Diffusion learner similar to Midjourney or StableDiffusion (text-to-image). More details on this module coming soon!

With this system, a researcher studying a disease or health condition caused by a certain molecule or protein could point this AI tool at that target. The module then generates a custom molecule that fits just in that target.This model however, is used to generate molecules!

This diagram gives an overview of the module flow. The molecules created by RFDiffusion are then fed into a system called LigandMPNN. This is a deep learning system that measures the level of interaction between the generated protein/molecule and a target molecule. LigandMPNN stands for 'Message-Passing-Neuralnetwork'. The result is passed to 'ColabFold' to filter the best results from LigandMPNN. Then the module re-generates a whole 'family' of proteins based on base hit in the last generation.

For example, here’s a protein the system came up with that could be a heart medication.

Learn more about Diffusion models for protein structure generation (and design) from the Rosetta Commons RaMP Bootcamp!

⚡ Developer Events & DX Updates

Run a Lilypad node using Docker

To improve the Resource Provider experience, our team has released a Docker RP option for the community! This guide is designed to help set up and manage a Lilypad Resource Provider using Docker, making it simple to deploy and maintain the node.

Follow these steps to contribute to the network, ensuring that your system is ready to handle the demands of modern AI workloads.

Weekly AMA Sessions!

To help answer questions and gain insights from our team, Lilypad is hosting AMA sessions every Wednesday. Catch up on all the latest developments from this week’s AMA with Alex, Logan, and Brian from the Lilypad Core engineering team! They discuss the Lilypad open source initiative, the latest developments at the network, and more.

We will alternate the timing of the AMA every week to ensure we can cater to RPs and devs across the globe! Find an AMA time that works for you on the Lilypad Calendar.

🌟Community

Pondside chat with DappaDan

In a new series focused on interesting chats with members of the community, the first Pondside chat was a huge hit! Steve from team Lilypad had an engaging conversation with the fabulous DappaDan.

Open source initiative!

We’re thrilled to announce the launch of Lilypad’s open source initiative, a significant step toward building a more collaborative, innovative, and inclusive tech ecosystem within the Lilypad Network community.

As the Lilypad Network is an open-source project, we encourage the community to contribute to (and Star!) our public GitHub repos:

• Lilypad - This is the core repo of the Lilypad Network, containing the code that powers Lilypad’s AI and compute platform.

• awesome-Lilypad - A curated list of resources, tools, tutorials, and examples related to the Lilypad Network.

• WindowsResourceProvider app - This repo contains the Windows Resource Provider app (beta), a tool designed to enable Windows users to run a node and contribute computational resources to the Lilypad Network.

• Lilypad documentation - This repo contains all the essential guides, best practices, and reference materials needed to work with the Lilypad Network. It's the primary resource for developers, resource providers, and users to navigate and contribute to the Lilypad ecosystem.

• Lilypad tools - A collection of tools and utilities to assist with various Lilypad-related tasks. Currently, it includes a Proxmox setup guide for managing virtualization and a script to update Lilypad. New tooling suggestions are appreciated!

Meme Contest!!

The Lilypad Meme-a-thon is in full swing! Don’t forget to submit your best memes by September 21 for a change to win one of the prizes.

Winners will have the glory of their memes showcased on our official social media channels! It's time to unleash your inner meme lord and make a splash in the pond.

The Guidelines to Leap In:

• Format: JPEG, PNG, or GIF. Keep it snappy!

• Deadline: All entries must be in by 9/21/24. Don't be late to the pond party!

• Submission: Submit your masterpieces through the Galxe Quest. Winners will be selected and notified after the quest end date.

• The Prize: Fame and Fortune: Winners will snag either:
  1st Prize: $250
  2nd Prize: $150
  3rd Prize: $100

🔮 What's Next?

To meet the expanding scope of the Lilypad network, our team has grown significantly. The engineering pipeline is brimming with updates aimed at enhancing job execution, refining the Proof-of-Work mechanism, and scaling the network overall.

Furthermore, our engineering and product leads are actively incorporating insights from the IncentiveNet into the development of the Lilypad mainnet. As we finalize the Whitepaper, the economic model, architecture, and other key aspects are being refined based on these valuable learnings.

☎️ Contact Us

💬 Have more questions or looking for guidance? Join our community AMA hosted every Wednesday!

Chat to us on Discord and follow along on X for the latest news and updates!