So this is what actually happened with $TAO / inside Bittensor:
Const (Bittensor’s father) called out Sam (the owner of the three subnets) for burning miners’ emissions, especially considering his subnets are among the top in emissions. Then Const sold a small but still significant portion of his SN3 tokens (which led to a drop in emissions) and also brought up an idea he’d been wanting to implement for a while, one that would give token holders more power. All of this made Sam realize that in the future it would be harder to pull off an exit scam by dumping all his tokens at once (especially after the feature introducing a delay on token sales would be implemented). That pushed him to take advantage of the moment, blame Const, and execute an $11M exit scam.
It's funny how DeepSeek went viral and https://t.co/Unn1UnGvmC was ready before it happened.
One week before this post. Chutes already had DeepSeek v3 running and usable for anyone over the world.
Bittensor subnet 64 🥩
EVM compatibility on Bittensor is a game changer. Developers can deploy smart contracts using tools like Solidity, Truffle + Metamask.
Bittensor bridging AI and traditional DeFi in new ways.
Expect to see more user-facing apps and staking protocols.
Think staking pools, liquid staking, and beyond. Developer innovation expanded.
Read More: https://t.co/j3811UbdDe
#EVM #Bittensor
Bittensor is incentivizing miners like @TensorplexLabs to train better foundational models through SN9, the pretraining subnet.
As a result, a 7b model is outperforming Llama 7b, Llama2 7b, and Falcon 7b on the benchmarks below.
It works, and it's just getting started.
$TAO
Bittensor is growing up and moving forward. The only constant was always change and (r)evolution.
Where do we go from here ... a 🧵about @taostats, weights, emissions, commissions and subnets.
[... > ]
almost done rewriting btcli to make the commands 90% faster.
current stats:
btcli s metagraph = 23s -> 2.2s
btcli s list = 4.5s -> 1.1s
btcli s get = 4.5s -> .7s
btcli s hyperparameters = 4.5s -> .7s
once finished, we will opensource the fork so that you can use it too. #TAO
"Corcel Chat is just an OpenAI wrapper"
Yeah right. Choose your model on @Corcel_X chat and API from the biggest and best centralised and open-source models.
#Bittensor
$TAO lists on @cryptocom 🔥
https://t.co/swN5W1cdvn
"https://t.co/GkT1BbM98Z App users can now purchase $TAO and other crypto easily with 20+ fiat currencies, and spend it at over 80 million merchants globally using the https://t.co/GkT1BbM98Z Visa Card"
The dominoes are beginning to fall..
$TAO #Bittensor
$TAO to list on @binance 🔥
#Bittensor will now be listed on the largest cryptocurrency exchange in the world
A huge effort from certain community members to make this happen
"Wen Binance.. Wen Tier 1?"
Now 🔥
$TAO
@marcoseveri93 @0xarrash A major cex listing will not solve this when the people who hold large amount of $tao isn’t selling. If they didn’t sell on the way to here and didn’t sell at 700+ or the way down to here I doubt they will when a major cex comes.
🚨 Highlights from Scaling $TAO Spaces:
The Industrial Revolution of AI is happening NOW.
Bittensor is the Holy Grail in AI because machine learning models are open ownership and distributed.
About exchange listings: Getting $TAO on exchanges isn't important - liquidity is already great, but shielded by fact that other projects pay @coinmarketcap etc. to inject false liquidity into markets. Don't want price manipulation, but authenticity for the purpose of longevity.
We will provide as much info. as possible to exchanges, audits, legal names, etc. But not actively pursuing listings. Want them to come to us. AND THEY ARE.
Why Bittensor will continue to grow rapidly: We don't need $7 trillion to work - we can optimize and be more efficient. No deals with Reddit, Twitter, no contracts. Buying GPUs will prove to be an inefficient way to organize capital.
Bittensor is a market mechanism - we only pay for what we want. So we can allocate productivity and optimize digital commodities.
2022: Single subnet
2023: Opened up to other teams to contribute, innovate
2024: Growing towards 1000+ subnets
It's technology can plug into data centers of @OpenAI
It can seep in through the cracks and grow far beyond what any company can ever raise.
Subnet 9 today released a model that is 40% better than OpenAI on Wikitext. We don't know who made it! Paid the best price we could for the lowest loss.
Meta layer on top of all world's computation - we can compete with the best in the world for the lowest price.
Subnet builders are the best in the world for fine tuning language models - and they can't compete with their own subnet. People are coming out of the woodwork and improving the models.
There is no way for any company or organization to compete with aggregation of data: Bittensor has added 10 terrabytes in a few months - 300 Gb added per day.
Not yet connected with pre-training, but that is coming soon.
Can provide numerous digital commodities at the best possible rates.
Subnet 27: can gather GPUs, drive the price down for users - maximizes value for holders.
By this year, could reach 1024 incentive mechanisms to drive value back to $TAO holders. Won't have unlimited subnets, but it's possible to scale much more.
Striving to be the neural internet of intelligence.
Next steps:
We are at the precipice of a huge exponential drive in what Bittensor can achieve.
Need to remove the root network - build a competitive network where thousands of subnets are interconnected and competing.
Dynamic $TAO: in 2-3 weeks @opentensor will have a proposal ready. Everyone will have a say.
Mobile wallet is coming soon!
1.5 year timeframe to move to PoS - specific to running chain nodes - sending weight transactions. After last week's issues, aren't ready yet. Want to focus on economic incentivization and building out the subnets.
(Chain is currently PoA - blocks run consensus mechanisms of subnets. Inside subnets it is currently a PoW/PoS hybrid)
Economic decentralization: Want to ensure that token itself is decentralized - that's why don't want root centralization, and why there is a tail for subnets and no pre-mine. Want a wide distribution of holders.
Bittensor's priority is to get the tokens into as many people's hands as possible to prevent issues ETH has faced 👇
Comments on Ethereum: ETH in contrast did a 50%+ pre-mine - end of day much of stake controlled by small group. Despite good infrastructure decentralization, don't have good economic decentralization.
@const_reborn was dissapointed that @VitalikButerin supports worldcoin, and doesn't think he fully understands the concept. Shared limited view of AI in crypto. Changed his mind somewhat after awhile.
Government pressure: governments love a narrative, will eventually be pushback regarding ability to train ML models - but we can move beyond whatever borders they create. Haven't seen any real pushback yet.
They know development is critical to expansion of their economies.
Overarching vision: creation of hundreds of teams - a nation-state that is collaborating and growing the ecosystem. They are building the best in-class tools for them.
Subnets are interoperable, but needs to be built by validators.
Slowly transitioning into more competitive systems, where validators will want you to purchase TAO before they give you access - not within the next 6 months.
Let me know if you have any questions!