Vinkkikeskus
Cryptoverse
Treidausareena
Cryptopedia
Lahja-arvonta
Eco Hub
The $ASP Buildkey Pre-TGE Trading is live on Aspecta platfrom
discoveryBanners
Trendaavat aiheet
#
Bonk Eco continues to show strength amid $USELESS rally
Hosico
Hosico+31,38 %
USELESS
USELESS+25,78 %
IKUN
IKUN+13,29 %
gib
gib+13,46 %
#
Pump.fun to raise $1B token sale, traders speculating on airdrop
Bonk
Bonk+7,2 %
ALON
ALON−4,6 %
LAUNCHCOIN
LAUNCHCOIN−15,2 %
GOONC
GOONC−17,29 %
KLED
KLED−8,24 %
#
Boop.Fun leading the way with a new launchpad on Solana.
BOOP
BOOP+1,42 %
Boopa
Boopa+1,4 %
PORK
PORK+1,8 %
header
samsja
leading research at @PrimeIntellect
samsja kirjasi uudelleen
will brown
will brown20.7. klo 18.32
i’m much more inclined to say that the RL *system* inside OpenAI is AGI rather than than any fixed model checkpoint which comes out of it
385,5K
samsja
samsja19.7. klo 23.25
I realized at our Berlin event that there are a lot of talented and ambitious young ppl in Europe. Just (almost) no inspiring company to build the future nor VC that have the balls to give them a chance. No wonder why everybody wants to come to sf|
Lazarz
Lazarz19.7. klo 00.41
Berlin I’m in you
44,8K
samsja
samsja19.7. klo 23.13
Open ai will be remembered as one of the most inspiring companies of all time
Noam Brown
Noam Brown19.7. klo 15.52
Today, we at @OpenAI achieved a milestone that many considered years away: gold medal-level performance on the 2025 IMO with a general reasoning LLM—under the same time limits as humans, without tools. As remarkable as that sounds, it’s even more significant than the headline 🧵
8,37K
samsja
samsja17.7. klo 17.27
I don't see the point of codebase defining batch size at a per gpu level. This means need to change the batch size param manually when up/down scaling experiment. I guess historically done in codebase that don't have grad acc ?
1,6K
samsja
samsja17.7. klo 00.06
We should be more worried about thinky hiring all the pytorch ppl than zuck poaching from open ai
14,08K
samsja kirjasi uudelleen
Max Ryabinin
Max Ryabinin16.7. klo 07.34
If you're at ICML and interested in verifiable inference, make sure to stop by our poster! We will present TOPLOC, an efficient activation hashing method that works across a variety of settings, e.g. switching inference setups or even models. July 16, 4:30pm, E-1106
1,41K
samsja kirjasi uudelleen
Benjamin Clavié
Benjamin Clavié15.7. klo 19.46
New blog post & new library are out now! The BP is about MaxSim, why it's *orders of magnitude* much more demanding than normal cosine similarity, and why GPUs don't care, but CPUs do! The library is maxsim-cpu, which makes it so CPUs can be fast and play it cool, too.
16,92K
samsja
samsja11.7. klo 22.58
absolute @kellerjordan0 victory
clem 🤗
clem 🤗11.7. klo 22.55
1T parameters, open-weights, just released on @huggingface!
10,28K
samsja
samsja11.7. klo 03.09
Curious to try this with diloco, would still do bs=1 on the inner optimizer and still get benefits of data parallelism
Micah Goldblum
Micah Goldblum10.7. klo 22.12
🚨 Did you know that small-batch vanilla SGD without momentum (i.e. the first optimizer you learn about in intro ML) is virtually as fast as AdamW for LLM pretraining on a per-FLOP basis? 📜 1/n
1,93K
samsja
samsja5.7. klo 22.29
also seniority in engineering is being able to work with tech debt of other
samsja
samsja5.7. klo 21.13
Unpopular opinion but technical debt is good, it's needed to accelerate and shrink timeline
2,34K