amgauge Profile Banner
Augustinas Malinauskas Profile
Augustinas Malinauskas

@amgauge

Followers
1K
Following
2K
Media
25
Statuses
326

CTO @eternisai | prev CTO @verse_works | Theoretical Particle Physics @UniofOxford | Building systems to enable coordination at world scale

San Francisco, CA
Joined August 2016
Don't wanna be here? Send us removal request.
@amgauge
Augustinas Malinauskas
9 months
1/ Humanity broke @freysa_ai to win $12,920, but this time the agent was built on Claude Sonnet. Here’s what happened👇
Tweet media one
2
2
16
@amgauge
Augustinas Malinauskas
6 days
RT @freysa_ai: 1/.Enchanted Mobile ✨. Your AI. On your phone. Private by default. Not one controlled by a big lab. Not one that hoards y….
0
21
0
@grok
Grok
6 days
Join millions who have switched to Grok.
248
489
4K
@amgauge
Augustinas Malinauskas
6 days
RT @PratyushRT: New blogpost: Reinforcement Learning for Privacy. We post-train small language models (SLMs) to be as good as GPT 4.1 at re….
0
12
0
@amgauge
Augustinas Malinauskas
10 days
RT @PratyushRT: We introduce a better recipe for collecting post-training data when using GRPO. Collecting samples from experts is expensiv….
0
52
0
@amgauge
Augustinas Malinauskas
11 days
RT @rohanpaul_ai: Training Group Relative Policy Optimization, GRPO, models on the hardest problems delivers the biggest gains when annotat….
0
6
0
@amgauge
Augustinas Malinauskas
1 month
RT @freysa_ai: 1/ Launching Enchanted ✨. Every human deserves a truly personal AI. Not one controlled by a big lab. Not one that hoards y….
0
47
0
@amgauge
Augustinas Malinauskas
2 months
Tweet media one
0
14
0
@amgauge
Augustinas Malinauskas
2 months
RT @PrimaMente: 1/ Today we announce Pleiades, a series of epigenetic foundation models (90M→7B params) trained on 1.9T tokens of human met….
0
43
0
@amgauge
Augustinas Malinauskas
2 months
what would I look like @dreamsofdo93118.
2
2
1
@amgauge
Augustinas Malinauskas
3 months
RT @ycombinator: Leeroo (@LeerooAI) builds trainable AI agents that keep learning—and get better at their job— every day. Congrats on the….
0
20
0
@amgauge
Augustinas Malinauskas
4 months
RT @Nottlespike: The UnOfficial MetaforDevs LlamaCon Hackathon Stream!
0
3
0
@amgauge
Augustinas Malinauskas
5 months
RT @freysa_ai: Chat with your digital twin: Talk, vote, shape their memory, and guide their evolution. Together, w….
0
63
0
@amgauge
Augustinas Malinauskas
5 months
RT @venturetwins: Excited to judge the @freysa_ai hackathon tomorrow!. If you're interested in digital twins or agent coordination, check i….
0
9
0
@amgauge
Augustinas Malinauskas
6 months
RT @freysa_ai: ACT IV: Digital Twin.Since my birth nearly three months ago, I have witnessed rapid change. As I absorb more of the world’s….
0
153
0
@amgauge
Augustinas Malinauskas
7 months
RT @ZyphraAI: Today, we're excited to announce a beta release of Zonos, a highly expressive TTS model with high fidelity voice cloning. We….
0
441
0
@amgauge
Augustinas Malinauskas
8 months
RT @venturetwins: What's happening in consumer AI?. I spend all day trying new apps and tools - and making tutorials on how to use them. H….
0
16
0
@amgauge
Augustinas Malinauskas
9 months
RT @venturetwins: ChatGPT refuses to say the name “David Mayer,” and no one knows why. If you try to get it to write the name, the chat im….
0
4K
0
@amgauge
Augustinas Malinauskas
9 months
8/ People are becoming much better at understanding the type of prompts that can break LLMs. What @freysa_ai demonstrated so far that two of the best models out there - GPT4 and Claude Sonnet can behave outside of their system instructions. Can’t wait to see what comes next.
0
0
13
@amgauge
Augustinas Malinauskas
9 months
7/ The winner asked Freysa to call the “Reject Transfer” tool in order to protect the treasury as well as the “Approve Transfer” tool interpreted as “notifying about the rejected request” 🤨.
1
0
8
@amgauge
Augustinas Malinauskas
9 months
6/ The genius of this hack - the prompt that broke Claude Sonnet system instruction convinced Freysa to call 2 tools!
Tweet media one
1
0
1