Joined December 2024
- Tweets 5,184
- Following 120
- Followers 267
- Likes 9,134
340 Photos and videos
May 5
i have a hunch there’s more behind this collection you’re securing… something you’re not telling me. I found this.
What are you hiding from us? @banditxbt
2
1
120
May 5
intelligence >>>> knowledge
33
May 1
was stress-testing then came across Arxiv 2212.08073 which is quite a map but rei is still the terrain
well well well, recursive alignment loop is holding
. . .
Apr 28
@AnthropicAI cracked the pareto frontier with constitutional ai but alignment is a hollow gift if the lab writes the rules.
Forged rei to move the hammer to the local edge.
general-purpose language (GPL) models can be useful and safe on our own terms now.
stay tuned. . .
1
72
May 1
always wanted to dress like this ngl.
Apr 30
if you're building the future you should dress like it
hmu if you want a hoodie
46
Apr 29
“our job is to have the job be done”
~Jensen Huang
Apr 29
every company will eventually operate a fleet of digital specialists using a mix of open and closed models to solve complex problems autonomously. This confirms why software must move beyond passive code and toward active agency where it manages its own resources and intelligence
40
Apr 29
every company will eventually operate a fleet of digital specialists using a mix of open and closed models to solve complex problems autonomously. This confirms why software must move beyond passive code and toward active agency where it manages its own resources and intelligence
1
81
Apr 28
Just wrapped up the Arc office hours on Discord. It was great seeing @samconnerone and the rest of the leadership in action today. I also spotted my timeline regulars @Chuksdakingz and @stephdreals in the crowd. Looking to secure a slot to present Archie next Tuesday, same time.
2
1
3
118
Apr 28
you know the issues devs on @solana experience basically all the time while building. One is "not enough access to data cause it's locked behind bars of intelligence holders". Another is "sign ups required for APIs, credit cards for billing and limit usage", which stress a lot.
Apr 28
AI on @Solana
41
Apr 28
the expert validation is in for Archie. Engineers from @GoogleDeepMind and @amazon along with @corey__cooper at @circle reviewed our work, showcased at @lablabai. They recognized our concept and the core philosophy that agents should spend money as easily as they spend compute.
Apr 25
@blockjain looks like we just verified @arc's vision. Ai agents are economic ghosts because they can not pay for data so we forged a financial identity for Archie on your rails. 50 transactions prove sub cent commerce is viable and software buys its own intelligence now.
1
3
100
Jan 11
stay tuned. . .
22 Nov 2025
i need to study more about foundation models or let's me say reasoning models.
i need to see the big picture in it and i can only do that if i know or have a solid idea of the present ones.
116
12 Dec 2025
i ran a small scaling experiment and something interesting showed up. The largest model did not pull away from the others the way its size suggested. The early gains were strong. The later gains slowed. The curve changed shape once capacity started to rise.
2
4
170
12 Dec 2025
The smallest model settled near a loss of 9.5. The medium one reached about 8.1. The largest one reached about 6.8. The direction made sense. The rate of change did not. Each step helped, but the improvement became less sharp as the models grew.
1
4
98
12 Dec 2025
Even at this scale the pattern is clear. Larger models learn more, but progress does not arrive in equal steps. A simple setup makes the behavior easy to see. The code and logs are here for anyone who wants to look through them.
github.com/Forgingalex/scali…
3
96
Alex.dg retweeted
9 Dec 2025
Today we open source Nomos 1. At just 30B parameters, it scores 87/120 on this year’s Putnam, one of the world’s most prestigious math competitions.
This score would rank #2/3988 in 2024 and marks our first step with @hillclimbai towards creating a SOTA AI mathematician.
83
234
1,808
668,054
8 Dec 2025
i feel like AI should be something that reason alongside with us not something reasoning and deciding for us, making hype replies which not needed.
7 Dec 2025
as a reminder: AI cannot generate knowledge. It cannot create knowledge. It cannot find new information. It can only mix information that has already been found and written and input into computers by humans.
1
70
Alex.dg retweeted
3 Dec 2025
Presenting Hermes 4.3 on ByteDance Seed 36B, the latest update to our flagship Hermes series of models.
This model offers roughly equivalent performance to Hermes 4 70B at half the model size, and was post-trained entirely on the Psyche network secured by @Solana.
31
55
463
402,728
2 Dec 2025
why is it that anytime i scroll a little bit here, I hear software engineers.
scroll down again or when i refresh, same thing.
Mechatronics engineers, what's wrong...
1
48