yusufozkan

4 ms·

1.
A single firm is behind OpenAI, Anthropic, and Meta hacking scandals(effort.news)
512 points by yusufozkan 19d ago|175 comments
2.
by yusufozkan 1mo ago
what the hell is that score/cost curve lol
3.
Reasoning walkthroughs for the ten problems [pdf](cdn.openai.com)
3 points by yusufozkan 2mo ago|0 comments
4.
What's the largest software project AI can complete on its own?(epoch.ai)
106 points by yusufozkan 2mo ago|106 comments
5.
by yusufozkan 2mo ago
> arc-agi-3 30.2% wow
6.
by yusufozkan 3mo ago
What is your definition of "actually intelligent"? I believe LLM's are more intelligent than the average human in a lot of ways according to the Legg/Hutter definition of intelligence: "Intelligence measures an agen
7.
Situational Awareness: The Decade Ahead(situational-awareness.ai)
3 points by yusufozkan 4mo ago|0 comments
8.
by yusufozkan 5mo ago
"This is a general-purpose LLM. It wasn’t targeted at this problem or even at mathematics. Also, it’s not a scaffold. We have not pushed this model to the limit on open problems. Our focus is to get it out quickly so that everyone can
9.
by yusufozkan 5mo ago
"The proof came from a general-purpose reasoning model, not a system built specifically to solve math problems or this problem in particular, and represents an important milestone for the math and AI communities."
10.
A New Era of Security: Frontier AI Defense(paloaltonetworks.com)
1 points by yusufozkan 5mo ago|0 comments
11.
New in Claude Managed Agents: dreaming, outcomes, and multiagent orchestration(claude.com)
4 points by yusufozkan 5mo ago|0 comments
12.
Talkie: An LM from 1930(talkie-lm.com)
4 points by yusufozkan 5mo ago|0 comments
13.
by yusufozkan 6mo ago
but people here had told me llms just predict the next word
14.
ARC-AGI-3 Leaderboard(arcprize.org)
5 points by yusufozkan 6mo ago|0 comments
15.
Harness design for long-running application development(anthropic.com)
5 points by yusufozkan 6mo ago|1 comments
16.
A Ramsey-Style Problem on Hypergraphs(epoch.ai)
3 points by yusufozkan 6mo ago|0 comments
17.
by yusufozkan 7mo ago
This is the same company that started as a nonprofit dedicated to open AI safety research, then became a capped-profit entity, then effectively closed-source, then dropped the cap, and is now pursuing full for-profit conversion. Every singl
18.
Firstproof_oai [pdf](cdn.openai.com)
2 points by yusufozkan 8mo ago|0 comments
19.
1X World Model – From Video to Action: A New Way Robots Learn(1x.tech)
3 points by yusufozkan 9mo ago|0 comments
20.
by yusufozkan 1y ago
> while pre-training our Llama 4 Behemoth model using FP8 and 32K GPUs I thought they used a lot more GPUs to train frontier models (e.g. xAi training on 100k). Can someone explain why they are using so few?
21.
by yusufozkan 2y ago
This is another cool example from their blog https://imgur.com/a/Svfuuf5
22.
by yusufozkan 2y ago
Generating an image of a completely full glass of wine has been one of the popular limitations of image generators, the reason being neural networks struggling to generalise outside of their training data (there are almost no pictures on th
23.
by yusufozkan 2y ago
Are you sure you are using the new 4o image generation? https://imgur.com/a/wGkBa0v