Hi! :fire: This entire concept was imagined based on my own ideas, with the help of artificial intelligence to structure and deepen certain technical reflections. My intention here is not to propose a definitive solution, but rather to invite everyone to open their minds and reflect together on the…

[image] Naval (@naval) on X @naval Crypto is the currency of the AI.

I like many have been thinking on how to merge AI and web3. I agree with the benefits of the approach but there are a few areas where I am not convinced: large scale training LST is mostly limited by network bandwidth (and not by cpu/gpu). It is extremely inefficient to train across datacenter…

Thank you very much for your thoughtful and pertinent comment! :heart: You’ve raised crucial points, particularly on the challenges of large-scale training, hardware costs, network bandwidth, and real-time inference latencies in a decentralized environment. These concerns are essential for any AI-W…

Additional insights and technical perspectives 1. TensorOpera and Aethir: Distributed GPU solutions One of the key challenges we discussed is the need to decentralize access to high-end GPUs for large-scale training. The partnership between TensorOpera and Aethir introduces an innovative solution b…

Decentralized AI Ecosystem with JamChain [Screenshot 2024-10-22 at 13.32.31] PoI and PoC Nodes PoI (Proof of Inference) Nodes and PoC (Proof of Contribution) Nodes are the two types of nodes actively participating in the network to provide computing power, validate inferences, submit datasets, and…

[image] Cyphertux: … Problem 1: Large Scale Training (LST) and Network Bandwidth As you mentioned, large-scale training of language models is limited by network bandwidth, not by the CPU/GPU itself. This poses a significant challenge in a decentralized environment where machines are geographica…

Hi Cyphertux, thanks for all the sharing and the debate around the topic. Distro, a project from Nous Research, recently proposed a decentralised architecture for LLMs using common GPUs and less than 100 times the power needed in legacy systems. These bold affirmations change the way model training…

Building the Future of Decentralized AI with Jam

Cyphertux October 21, 2024, 1:34am 2

Topic		Replies	Views
Decentralised Futures Program: MatrixAI - A decentralized AI computing power protocol Ecosystem decentralized-future	0	505	November 24, 2023
Cybergov V0: Automating trust-verifiable LLM governance on Polkadot Tech Talk governance	6	408	September 10, 2025
Deitos Network: Forging a New Path in Big Data and AI with Blockchain Technology Ecosystem	3	729	December 1, 2023
ArcheLabs Progress Report: Laying the Infrastructure for AI on Polkadot Ecosystem	2	223	May 22, 2026
Decentralized Voices Cohort 5 - Light track - Karim/Cybergov Governance	9	499	September 6, 2025

Building the Future of Decentralized AI with Jam

Related topics