Bagel ??

Today, the Bakery opens its doors to all open source AI developers. Bagel makes open source AI monetizable. Our novel AI model architecture enables anyone to contribute while ensuring developers receive revenue attribution. The Bakery, the first product built on the Bagel architecture, revolutionizes how AI models are fine-tuned and monetized. With this launch, we're also unveiling our integration with @huggingface. Through this integration, you gain access to most cutting edge open source models like: Llama-3.3 for streamlined and efficient language capabilities. Qwen/QwQ for advanced language innovation. Stable Diffusion for next-generation image creation. This is the foundation for open source AI’s evolution. The future of monetizable open-source AI begins now. Step into it:?https://bakery.dev/

In this episode of Everything Bagel, co-hosts Alex Kehaya and Bidhan Roy, Founder of the Bagel Network, discuss the groundbreaking advancements in zero-knowledge proofs (ZKPs) and their impact on open-source AI monetization. Bidhan shares how Bagel Network is revolutionizing open-source AI by enabling developers to monetize their contributions while preserving ownership and privacy. They discuss the recent breakthrough with ZK LoRa, a zero-knowledge proof system for AI training, which drastically reduces verification times from weeks to under two seconds—a game-changer for decentralized AI. The conversation explores how community-driven ASI (Artificial Super Intelligence) can emerge through specialized AI models, making decentralized AGI (Artificial General Intelligence) more accessible and cost-effective. Alex and Bidhan also touch on how The Bakery, Bagel’s latest product, is helping developers build and stack modular AI contributions like LEGO pieces, allowing for greater scalability and efficiency.

Builders and bagels, what’s not to love? ?? We just brought together 50+ AI/ML researchers and devs in SF to swap ideas, demo fresh use cases, and discuss the future of monetizable open source AI. We’re grateful for all the insights, questions, and collaboration that sparked throughout the day. This is the time to build. Monetizable open-source AI is real.

Everyone’s asking how we pulled off #ZKLoRA, enabling zero-knowledge proofs for rapid LLM fine-tuning. In the latest episode of “Everything Bagel ??,” our founder Bidhan R. shares how our cryptographers & ML researchers tackled the toughest scalability and privacy challenges to set a new standard. Tune in to learn more, and check the comments for our open research & GitHub repo! The era of Monetizable Open Source AI is here, and it starts with Bagel ??.

Can open-source AI be the great equalizer? Exploring how democratized artificial superintelligence could reshape society, power dynamics & human purpose. From tech giants to underdogs like Mistral & DeepSeek - diving into the race for AGI, revenue models & ethical implications. New episodes of Everything Bagel drop every Wednesday! ??

ASI isn’t just next-level AI—it’s god-level intelligence. On the latest episode of Everything Bagel ??, our founder Bidhan R. breaks down why open-source AI is humanity’s best defense against centralized control. This episode dives into: - What ASI means for humanity’s future - The risks of centralized AI control - Why open-source is the only way forward - How we’re building monetizable open-source AI for builders The race to superintelligence is on, watch the full episode here: https://lnkd.in/eyWcBdfw #AI #AGI #ASI #OpenSource

"We want open-source AI to win. We want developers to build better models than top labs. But how do we make that happen?" Our founder Bidhan R. explains how Bagel’s ZKLoRA protocol and infrastructure make this vision reality. Link to full episode of Everything Bagel ?? in the comments

Join us for a special edition of Everything Bagel, where our founder, Bidhan Roy, will discuss the role of monetizable open-source AI in advancing AI development. We’ll get into how methodologies like Test Time Scaling and MoE, enable efficient model development and how the Bagel infrastructure is designed to empower contributors and accelerate progress toward community-driven AGI/ASI. See you there!

ZKLoRA: Zero-Knowledge Verification for LoRA Fine-Tuning is Here ?? In 2024, zero-knowledge verification in machine learning felt out of reach. Latency overheads made it impractical. But now it’s 2025. The landscape has shifted. And today, we are proud to release ZKLoRA—an open-source zero-knowledge protocol that verifies LoRA fine-tuning updates in just 1–2 seconds. Not for small models. For state-of-the-art open-source models like LLaMA 3.3, with tens of billions of parameters. How it works: - LoRA adapter weights stay private. Alignment with the base model is verified cryptographically. - Overhead? Minimal. Trust requirements? Eliminated. - Built for real-world workflows, with support for multi-party inference. Here’s what happens in practice: 1?? The base model user processes a forward pass locally. 2?? Partial data is sent to a LoRA contributor, who returns computed outputs. 3?? Cryptographic proofs validate the exchange. ZKLoRA merges LoRA’s parameter efficiency with zero-knowledge proofs. Low-rank adapters reduce complexity, while cryptographic checks ensure correctness. It’s a seamless union of cryptography and AI. ?? Explore it here on GitHub: https://lnkd.in/gdiZBujv We believe ZKLoRA paves the way for secure, trustless collaboration in decentralized AI.

2025 is the year of monetizable open source AI ?? Be part of our first Builder’s Day on Feb 8! We'll be joined by John Alioto from Google for Startups and 75+ builders. Space is limited. Register here: https://lu.ma/hyvbkv2e