Mutex::
@naschain_ai
Followers
373
Following
110
Media
30
Statuses
108
How is this hard to get, Alex? 1M GPUs are for running SOTA models the company develops. Investors invest in research and the technology not the ability to build clusters and give compute away for free open weight LLMs from investors pocket. GPU is infra for them not a product!!!.
Here's what people think about OpenAI 1 million GPUs. But in Bittensor, people complain that we have to fund compute for $TAO's success. I guess OpenAI investors can see further than the short-term pains of the initial investment required to make something great
0
0
0
Incentivize ‘Publish the Epstein list,’ and watch Congressmen and Congresswomen start mining.
1
0
8
Back in the day, we used to call this an "agent". Today, we string a few prompts together and dub it an “AI agent.” 🙃
1
1
1
Shopping for the validators. No more GPU validation. Only physical validation.
1
0
2
Your “state-of-the-art” depends entirely on someone else’s state-of-the-art. You are basically just a glorified import statement (weight loading) with delusions of grandeur.
0
0
0
He got the idea from AirTags and Apple devices, but he didn’t realize that while he has the AirTags, he doesn’t have the Apple devices. There are 2.2 billion of them scattered around the globe.
🇺🇸 BILLIONAIRE JACK DORSEY JUST LAUNCHED "BITCHAT," A NEW BLUETOOTH-BASED MESSAGING APP. NO WI-FI. NO CELL SERVICE. FULLY END-TO-END ENCRYPTED 👀
0
0
1
The most influential, craziest "Novelty Searches" happened pre-dTAO. The pressure to show progress was insane, and the bar was massively high. You couldn’t BS around; validators asked tough questions. Now it’s just a family get-together with a “meh, we’ll see” vibe.
0
0
0
If I had a dime for every time a new compute subnet was created, I could fund the next compute subnet .
1
0
7
Might be wrong but I don’t think convexity of the per-example score really matters here, because we’re still optimizing a convex overall objective.
Hi @tplr_ai. I read your wp and its technically almost perfect. I did notice one thing though. Your score is the difference of losses but technically the difference of two losses is not convex. I would like to humbly note that a softmax score:. exp(−ηℓi(θt+1))
1
0
0
RT @svlevine: If you have a policy that uses diffusion/flow (e.g. diffusion VLA), you can run RL where the actor chooses the noise, which i….
0
156
0
In reality, it boils down to element-wise subtraction of two vectors. Be aware!.
Introducing MANTIS! The Ultimate Signal Machine running on netuid 123. MANTIS scores miners based on the information theoretical value of their embeddings to the prediction of BTCUSD. Incentives are distributed proprtional to a given miners contribution to the accuracy of a
1
0
3
The math was in the proposal. We accepted it, backed it, implemented it. ALPHA holders did 2x, 4x, 10x in weeks. BTC ↓ → $TAO ↓ → ALPHA ↓ → Investor confidence ↓. Rewriting the incentive every time your bag is in red is ridiculous. First, it's 0% root, next is alpha *= 2.
0
0
3
The strongest crypto asset, Bitcoin, is bleeding, while the guy with the 'Chutes' bag expects it to outperform Bitcoin and behave like a stablecoin. However, the root is the problem, not WW3. .
3
0
6
You bought alpha with your own conviction. Others didn’t. Now you’re down 50% and blaming root.trying to get others to swap their TAO so you can dump and exit . Set Root APY to zero, nothing changes. Let subnet quality bring investors and pump your bags, not your crying and blame.
0
0
5
People are dying on both sides. Millions are losing sleep in fear. And this absolute fraudster is worried about his TAOs—next thing you know, he’ll be announcing a 'Peace in the Middle East' subnet.
0
1
4
Bring back the power to validators to control the emissions and kill all crypto mining SNs before its too late and they have taken over the network.
2
0
2