Trendaavat aiheet
#
Bonk Eco continues to show strength amid $USELESS rally
#
Pump.fun to raise $1B token sale, traders speculating on airdrop
#
Boop.Fun leading the way with a new launchpad on Solana.
Macrocosmos
SN1: faster, cleaner, sharper 💥
✍️ Entirely rewritten the SN1 codebase, making it 95% smaller, increasing speed and optimization
🔎 Clear pathway to releasing Deep Researcher
🔥 Added a secure sandbox to enable our agents to execute their code - our Fast Incentivized Reasoning Engine (FIRE) is available through the app and API
584
Biggest month yet? 👀 Check out our subnet status report for June's highlights:
🌐 SN9’s distributed pretraining architecture IOTA launched
🖼️ #NoveltySearch in The Louvre @proofoftalk
⚡ SN1 GAN results
🚀 SN13 gets SDK + payments go live
📰 Macrocosmos featured in @Forbes
See for yourself - report in our thread🧵
10,36K
Thanks @VenturaLabs for hosting us - watch @macrocrux and @Felix_Quinque discussing how we built IOTA on #SN9
30.6.2025
Ventura Labs Ep. 50 - Steffen Cruz and Felix Quinque
Felix Quinque, AI Lead, and Steffen Cruz (@macrocrux), CTO & Co-Founder Macrocosmos (@MacrocosmosAI)
Timestamps:
1:24 – Introduction
2:12 – Subnet 9’s Transition to IOTA
2:47 – The Problem: Large Language Model Training
4:51 – Layer Groups Overview
5:02 – Defining a Layer in an LLM
6:07 – Layer vs. Dataset Partitioning
7:21 – Data Distribution Among Miners
10:08 – Introducing Pipeline Parallelism
11:18 – Orchestrator: Scheduling & Coordination
15:44 – Validation: Shadow Audits
19:39 – Future of the Validation Process
24:51 – Activation Compression Importance
31:30 – Butterfly All-Reduce Explained
36:43 – Proudest Achievements of Subnet 9
39:33 – Defining Success for IOTA Subnet
46:21 – Advice for Zero-to-One Teams
52:16 – Conclusion & Thanks
4,89K
Johtavat
Rankkaus
Suosikit