Macrocosmos

896 posts

Macrocosmos banner
Macrocosmos

Macrocosmos

@MacrocosmosAI

Building distributed intelligence on Bittensor. SN1, 9, 13 https://t.co/LizGMGPFh4 https://t.co/weDmtZxH4V

London Katılım Mart 2024
25 Takip Edilen6.4K Takipçiler
Sabitlenmiş Tweet
Macrocosmos
Macrocosmos@MacrocosmosAI·
Today, we are announcing the launch of Training at Home, our effort to democratise the creation of AI models to allow the many to participate, not just the few. Training at Home is an easy to use application that allows anyone with a Macbook to contribute to creating AI models. No complex coding, no complicated set-up - just install and train. This is the next step of iota’s mission to scale distributed training until it is indistinguishable from training in a data centre. Together, we can train models collectively, and you can help us shape a more open, democratic, and accessible future for AI. Contribute, compute, own the future. Coming soon.
Macrocosmos tweet media
English
39
64
313
71.5K
Macrocosmos
Macrocosmos@MacrocosmosAI·
An important next direction is to validate these gains at larger parameter scales, across broader architecture families and downstream tasks. Our preliminary experiments suggest that ResBM can achieve similar performance in these cases, but we leave this to future work. Read our paper on arXiv here. arxiv.org/abs/2604.11947
English
0
0
15
404
Macrocosmos
Macrocosmos@MacrocosmosAI·
The ResBM architecture establishes that we can train end-to-end with no degradation in convergence rate relative to uncompressed baselines. These results suggest that pipeline-parallel pretraining is no longer fundamentally limited by inter-stage communication, and that architecture-level design for communication efficiency is a practical path to scaling training to distributed compute networks.
Macrocosmos tweet media
English
1
1
18
769
Macrocosmos
Macrocosmos@MacrocosmosAI·
Today, we present ResBM (arxiv.org/pdf/2604.11947), a 128x activation compression technique for achieving SOTA training results in low-bandwidth, distributed communication settings for pipeline parallel training across the internet. This technology underpins @IOTA_SN9 - our distributed training platform.
Macrocosmos tweet media
English
9
36
152
20.6K
Macrocosmos
Macrocosmos@MacrocosmosAI·
We spoke on the @twistartups podcast, sharing our perspectives on Bittensor, along with our three subnets: @Apex_SN1, @IOTA_SN9, and @Data_SN13. See the conversation here. Timestamp: 42:35 mins
This Week in Startups@twistartups

A subnet founder allegedly walked away with $10M in TAO and torched his own community in the process. On TWiST, Mark Jeffrey of Stillcore Capital joins us to break down what really happened with Templar/Covenant, the overall fixes that co-founder Const has proposed, and why Bittensor’s incentive engine may have been a victim of its own success. PLUS we’re joined by subnet operators Will Squires and Steffan Cruz (of MacroCosmos) and Ken Miyachi of BitMind to get their perspective on the controversy, and to demo the exciting projects they’re still building on the blockchain. 0:00 Mark Jeffrey joins the show! x.com/markjeffrey 2:18 How Mark Jeffrey learned about Bittensor. bittensor.com 6:17 Plaud: If your work depends on conversations — interviews, meetings, calls — you need a Plaud NotePin. You can check it out at Plaud.ai/twist and use code TWIST for 10% off! 7:22 Mark Jeffrey's Bittensor investments. stillcorecapital.com 9:25 Check out our discussion with Nova: youtu.be/gjRt4eUyiYc?si… 10:16 Sentry - New users can get $240 in free credits when they go to sentry.io/twist and use the code TWIST 10:41 Check out Ridges! ridges.ai 11:53 How trading alpha tokens works on Bittensor 12:44 Subnet drama: what happened? x.com/covenant_ai/st… 16:01 Do subnet owners have too much power? 18:33 Check out our conversation with Sam Dare (2268): youtu.be/TN2RmNuX4-k?si… 19:10 How Sam Dare should've handled walking away (per Mark Jeffrey) 20:02 Deel - Founders scale faster on Deel. Set up payroll for any country in minutes, hire anyone anywhere, get visas handled fast, and get back to building. Visit deel.com/twist to learn more. 23:29 Who should subnets be owned by? 24:02 Ken Miyachi from BitMind joins the show x.com/kenmiyachi 30:56 Netsuite - Get the free business guide Demystifying AI at netsuite.com/twist 31:06 Ken's $3M raise & investors (Arch, Canonical, Mechanism) 33:18 Token vs. equity: how to think about a subnet investment. 41:57 Will Squires and Stefan Kruse of MacroCosmos join the show x.com/willsquires x.com/sczsteffencruz 42:54 How MacroCosmos lets anyone become a compute provider. iota.macrocosmos.ai 56:29 Stefan on the Covenant drama: "disappointing, but solvable" 1:02:11 Off-duty with J-Cal, Mark Jeffrey, and Lon Harris 1:02:48 Bieber vs. Carpenter: does Coachella owe you a spectacle? youtube.com/watch?v=hp5O72… x.com/FashionXGirl_/… 1:15:20 Jason says Staples should pay the "Staples baddie" $1M/year @blivxx" target="_blank" rel="nofollow noopener">tiktok.com/@blivxx cc: @Jason, @Lons, @MarkJeffrey, @const_reborn, @DistStateAndMe, @tplr_ai, @Covenant_ai, @MacroCosmosAI, @Apex_SN1, @IOTA_SN9, @KenJon, @BitMind @BitMindAI 🎥 Watch the full episode here 👇

English
0
8
52
2K
Macrocosmos
Macrocosmos@MacrocosmosAI·
Training frontier models over the internet requires new techniques. Today, we present ResBM, a residual encoder-decoder bottleneck architecture that enables 128x activation compression for low-bandwidth distributed pipeline parallel training. Developed for @IOTA_SN9, we show SOTA compression without significant loss in convergence rates, increases in memory, or compute overhead. Expect the full paper release in the next 72 hours.
Macrocosmos tweet media
English
14
44
213
46.8K
Macrocosmos
Macrocosmos@MacrocosmosAI·
Bittensor always emerges stronger from pain. Hacks, exploits, disagreements all fade into the history of this protocol. The flame of distributed training continues to burn brightly. One individual does not make a nation. We stand with @const_reborn.
English
2
25
148
2.9K
Macrocosmos retweetledi
Apex・SN1
Apex・SN1@Apex_SN1·
We just made Apex (@Apex_SN1) agent-native. New CLI commands mean an AI agent can now discover competitions, read the docs, submit solutions, and check its own scores — completely autonomously. No human in the loop. Here's what shipped 🧵
Apex・SN1 tweet media
English
6
23
116
23.2K
Macrocosmos retweetledi
Data Universe ・ SN13
Data Universe ・ SN13@Data_SN13·
dataverse-cli is the fastest growing repo in Macrocosmos history.
Data Universe ・ SN13 tweet media
English
5
7
60
5.8K
Macrocosmos retweetledi
Data Universe ・ SN13
Data Universe ・ SN13@Data_SN13·
Introducing `dv` - a Rust CLI for querying real-time social data from X & Reddit. Powered by Bittensor SN13's decentralized miner network. ``` dv search x -k bitcoin -l 100 ``` One command. Live data. No middleman. Open source. Built for agents. 🧵👇
Data Universe ・ SN13 tweet media
English
15
53
656
64.9K
Macrocosmos retweetledi
Apex・SN1
Apex・SN1@Apex_SN1·
We’ve built a simulation of the @IOTA_SN9 communication network. This is a high-fidelity digital twin - an abstracted version of our distributed training architecture, designed as a testing ground to run experiments and develop novel algorithms to increase the speed and quality of model training. We’re using the simulator as an environment for open competitions on Apex, outsourcing algorithmic innovations to the Bittensor miners. It’s the first time our simulator can be interacted with by the public. Our opening simulator competition is live now.
English
3
12
77
19.7K
Macrocosmos retweetledi
Apex・SN1
Apex・SN1@Apex_SN1·
Apex and @IOTA_SN9 are working together again. The IOTA simulator competition launches later today. Join us as we accelerate distributed training.
English
7
24
122
15.3K
Macrocosmos retweetledi
Apex・SN1
Apex・SN1@Apex_SN1·
We recently concluded our matrix compression competitions on Apex. The purpose of these was to increase the training speed of @IOTA_SN9 by speeding up data transfer. We found that by outsourcing the development of matrix compression algorithms, Apex achieved a ~3x reduction in data transmission size. Our winning algorithm strictly improves latency for network speeds < 50 MB/s. See the full results in our Substack below
Apex・SN1 tweet media
English
2
12
79
8.8K
Macrocosmos retweetledi
Data Universe ・ SN13
Data Universe ・ SN13@Data_SN13·
Data Universe has reached 2,500+ subscribers. Our webscale social media data collection app is growing by the day. People from all sectors, not just #Bittensor, are harnessing subnet 13’s real-time stream of online information, perfect for #SocialListening and #MarketResearch. Try our no-code solution to affordable data yourself. The first $5 of credits are free
Data Universe ・ SN13 tweet media
English
1
8
72
5.5K
Macrocosmos
Macrocosmos@MacrocosmosAI·
Reminder: we’re going live tomorrow Two core members of our engineering team, @mccrinbc and @Felix_Quinque, will be talking about @IOTA_SN9 and Train at Home. A technical deep dive, breakdown of iota’s goals and milestones, and a Q&A giving the community a platform to be heard. 📍 Location: X livestream (on the @MacrocosmosAI X account) 📅 Date: Thursday 5th March 🕜 Time: 4pm UK time
English
2
7
19
3.1K
Macrocosmos
Macrocosmos@MacrocosmosAI·
Learn about @IOTA_SN9’s Train at Home - from the engineers who helped build it. This week, Inventive Mechanisms is going live with @mccrinbc (Brian McCrindle) and @Felix_Quinque (Felix Quinque), two of SN9’s core devs. This is a deep dive into Train at Home, looking at its architecture, technical setup, and future. If you’ve got any questions, this will be the time to ask them. 📍 Location: X livestream (on the @MacrocosmosAI X account) 📅 Date: Thursday 5th March 🕜 Time: 4pm UK time
Macrocosmos tweet media
English
2
3
22
3.4K

Keşfet

@IOTA_SN9 @twistartups @Apex_SN1 @Data_SN13 @const_reborn @mccrinbc @Felix_Quinque @elonmusk