Troy (@troyquasar) 's Twitter Profile
Troy

@troyquasar

Quasars are the brightest objects in the universe ✨

ID: 1805332450119225344

calendar_today24-06-2024 20:09:48

1,1K Tweet

205 Followers

1,1K Following

Ricardo Monti (@ricardomonti9) 's Twitter Profile Photo

1/ People often think better multilingual models must come at the cost of English performance. Not true. The constraint isn’t capacity, it’s data quality, and we can fix it. Today @datologyAI shares ÜberWeb: a year of multilingual curation lessons, scaled to 20T+ tokens.

1/ People often think better multilingual models must come at the cost of English performance. Not true. The constraint isn’t capacity, it’s data quality, and we can fix it.

Today @datologyAI shares ÜberWeb: a year of multilingual curation lessons, scaled to 20T+ tokens.
Algod (@algodtrading) 's Twitter Profile Photo

For bittensor to succeed the subnets quality have to drastically increase (hard task) A decent majority of subnets are a waste of money, no path to profitability, incentivise mechanism broken, overfitting benchmarks and pretending they beat SOTA Community has to be more

Ryan Sun (@sun_hanchi) 's Twitter Profile Photo

DSA’s inability to train from scratch made mid train harder GLM5 did 1T at 32k, 500B at 128k, and 50B at 200k context length Meanwhile, Qwen3-Next-Coder directly go for 1T at 262k due to GDN DS V4 might has sparse attn that trains from scratch, enabling long ctx all the way

Matt Karas (@the_mattyk) 's Twitter Profile Photo

You can now pay with ANY subnet alpha token on Targon. And we'll even give you a 10% discount if you do. Bittensor is stronger when alpha tokens are valued with real utility.

Autopoiesis (@autopoiesislab) 's Twitter Profile Photo

It's here. Aristotle is now live. AI built for how scientists actually think. Self-skeptical reasoning, epistemic graph exploration, bold hypotheses grounded after generation. Free for verified researchers in the United States. autopoiesis.science/blog/introduci…

Khala Research (@khalaresearch) 's Twitter Profile Photo

Training a frontier AI model cost $5M in 2020. Today it costs $1B+ The gap keeps compounding, in favour of the same five companies A decade ago, nobody believed open-source could compete with enterprise software. But then it did Bittensor (TAO) is a global open market where

Training a frontier AI model cost $5M in 2020. Today it costs $1B+

The gap keeps compounding, in favour of the same five companies

A decade ago, nobody believed open-source could compete with enterprise software. But then it did

Bittensor (TAO) is a global open market where
Quasar (@quasarmodels) 's Twitter Profile Photo

Big news: Quasar just signed its first enterprise contract. Ayles Flow (a canvas-agent interface, out of stealth today) is running a version of Quasar we built specifically for them. Long-term deal. Custom model. From day one. Every upgrade the miners push makes their product

Big news: Quasar just signed its first enterprise contract.

<a href="/aylesflow/">Ayles Flow</a> (a canvas-agent interface, out of stealth today) is running a version of Quasar we built specifically for them.

Long-term deal. Custom model. From day one.

Every upgrade the miners push makes their product
Punisher ττ (@cryptozpunisher) 's Twitter Profile Photo

Bittensor >> $TAO >> $dTAO Startup 24: Quasar First enterprise contract signed. Ayles Flow X Quasar No noise. No empty promises. A real client. Long-term deal. Custom model. From day one. Ayles Flow is already running on a dedicated version of Quasar built

Bittensor &gt;&gt; $TAO &gt;&gt; $dTAO

Startup 24: Quasar 

First enterprise contract signed.

<a href="/aylesflow/">Ayles Flow</a>  X <a href="/QuasarModels/">Quasar</a>

No noise.
No empty promises.
A real client. 
Long-term deal. 
Custom model.
 From day one.

Ayles Flow is already running on a dedicated version of Quasar built