🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

BτComander templar Scaling an AI model from 1.2 billion to 8 billion parameters is like expanding a football team from a small local squad to a world-class club. A bigger team can cover more ground, execute complex plays, and adapt better, just as more parameters help an AI tackle tougher tasks and

templar Not a number because the loss is so low bro. Its a good model bro. AGI achieved bro.

Just released a detailed deep dive on decentralized training. We cover a lot in there, but a quick brain dump while my thoughts are fresh: So much has happened in the past 3 months and it's hard not to get excited - Nous Research pre-trained a 15B model in a distributed fashion

Teng Yan · 30 days of COT Nous Research Chain of Thought — 🌈 30 Days of AI x Crypto Teng, the only one live on mainnet is sn3 templar templar -- all others you mention are on testnets. Great coverage tho!

templar (sn3 on bittensor) also falls into the data-parallel bucket and is already live. in their latest run, they're using a permissionless network to train an 8B parameter model with refined incentive structures to improve training coordination.

We've got you fam : soundcloud.com/tplr-ai/missio…

x.com/i/article/1931…

Mark Jeffrey templar Distributed State

8b now

templar ☺️

Internet's down? Don't care. The run's still going

With every run, miners improve exponentially. "We are yet to observe a task , that well incentivised miners could not accomplish" templar

_τao_moonwalker_ templar Hats off to the miners. Honestly , these guys are the SEALs of bittensor. We put them through the Gauntlet , and they always rise to the challenge.

Miners is GOD GOD is a miner. Or Carpenter Distributed State templar const

when we achieve 70B, fully decentralized, permissionless and opensource AI templar

The approach: Templar (SN3) → Base model pretraining Gradients (SN56) → Instruct fine-tuning Proof of concept results: Templar 3B (mid-training) → Gradients instruct tuning → benchmarks rising across the board Plot twist: we're just getting started.