Token Engineer, Founder :| ReFi R&D @ https://teapawt.net/t.co/q7SoviE9Ld | Stealth Mode Hybrid Fund | KERNEL Genesis Block | R&D @NATIXNetwork | @Kernel0x Fellow
Right in behind you
Joined January 2018
- Tweets5.3K
- Following687
- Followers409
- Likes1.9K
if @realDonaldTrump cared about crypto and his most loyal #maga at all - he would go hard after the hackers from North Korea who hacked @Bybit_Official
2
62
Aaasrgh, @gpt4all default embeddings device is the CPU not the GPU?
Bummed out. Definitely not fast enough for weekly updates when using CPU (ok, this computer has an ancient quad-core CPU)
1
115
ft.com/content/17606f25-1d03…
The number of young men in G7 countries who aren’t in school, working, or job hunting has been rising for decades. This trend is concerning—it could lead to a generation of broke, lonely men who are more likely to adopt extreme views. Fewer ...1/2
1
49
An unexpected moment of clarity in the author's thought process ... yet squandered on an elegy of web2 "good ole times"
Relevant screenshots below, use f.i. Brave to avoid paywall.
What does CT think about this? Are we before or after peak nostalgia?
nytimes.com/2024/09/06/world…
1
20
I have observed an interesting phenomenon: when comparing llama-3.1-405b and claude-3.5-opus AND sonnet (with a shitty unoptimized prompt), llama3.1 appears to interpret prompt repeating (replying with a copy of the initial prompt) as a cue to answer the OP in more detail! 1/2
1
117
This appeared to work up to depth 3.
OTOH @AnthropicAI Claude kept repeating the initial answer whenever the OP was repeated + far worse reply fidelity.
Anyone seen similar behaviors in other models? What prompt types achieve similar behavior?
Tested on arena.lmsys.org/
31
Beowulf clusters are going to make a comeback nobody expects
Just wait until a follow up project adds distributed vram
Woohoo gpu.cpp has now been officially released!
This is a really exciting project which provides a device-independent way to use GPU compute. No more writing separate code for CUDA, AMD, Mac, and Intel GPUs!
We can't wait to see what you all are able to create with this.
18
I agree with this shift in dominance - unless the preexisting infra becomes so accessible to the expected size of the global builders pool that "all" apps will i.e. bootstrap via community
23/23
As we shift from infra to consumer apps, funds will need to focus less on founders who build a practical PMF, and more on those building an experience people love.
Because its far easier to monetize something people love, than to make them love something well monetized.
27
We need use, sadly, capitalist approaches to correct the negative side effects of capitalism.
It's a marathon, it's a life task.
It's making the invisible externalities visible to the institutions that interact with them. By turning the free-rent tap off, we evolve capitalism.
1
52
@gumsays just wanted to let you this account was an insta follow for me.
Subverting the identity/ brand of stockgumshoe ... Priceless.
I want an NFT for.the occasion. Network of your choice 😄
5
Grabbed some more $NATIX on @CoinList. Can't have enough XD @NATIXNetwork #UpOnly offers.coinlist.co/o/natix-t…
1
1
30
I keep wondering when the @googlemaps airdrop is going to finally be announced. Users provided probably the majority of value: reviews and places that the team never would've been able to achieve no matter the cost
19
I wonder if the world would be a better place today if we had grown up to treat other's personal data the way we treat ionizing radiation.
1
1
59
Behaviour modification Behaviour modification Behaviour modification Behaviour modification Behaviour modification Behaviour modification Behaviour modification Behaviour modification Behaviour modification
6
42
