interstellarninja
@intrstllrninja
Followers
2K
Following
19K
Media
882
Statuses
6K
growing artificial societies | by the open-source AGI, for the people | building @MarketAgentsAI | github: https://t.co/ZCc6Pwfnbm
Tesseract
Joined December 2010
this interstellarninja is on covert missions right now involving power struggles with closed source AI labs and regulatory bodies plotting against open source AI 🥷.
Japan’s ninja are famed for their covert activities over centuries of power struggles in the country, and were highly prized by Tokugawa Ieyasu.
1
1
15
RT @MistralAI: Mistral Document AI is now in @MicrosoftAzure AI Foundry!. Ever struggled to extract useful info from PDFs, scans, or comple….
0
146
0
RT @leavittron: Very excited to announce BeyondWeb, @datologyAI’s synthetic pretraining data generation paradigm. BeyondWeb is a rephrasing….
0
40
0
RT @_Dave__White_: You spot an unsigned band destined to break out. Instead of cold-calling labels, what if you could bet on them yourself….
0
33
0
RT @osanseviero: Introducing Gemma 3 270M 🔥. 🤏A tiny model! Just 270 million parameters.🧠Very strong instruction following.🤖 Fine-tune in….
0
335
0
RT @bihanmahadewa: apple couldn’t make a car but a chinese phone company did. $30k, 500 mile range. 80% charge in 10 minutes. already 250….
0
5
0
ai agents playing games simulated by another ai unlock vast amounts of training data.
Demis Hassabis describes a surreal setup: one AI playing inside the mind of another. DeepMind SIMA agent can now operate within Genie 3 -- taking actions in a world that Genie generates in real time. "an AI creates the world, another acts inside it". this setup could unlock
0
0
3
ai agent swarms' potential remains unrealized.
Ants are dumb. but ant colonies are smart. More generally, there's the idea of swarm intelligence. Quite a lot of research in this area. Might be applied to crypto and AI agents specifically. And distributed systems more generally.
0
0
3
still bullish on agi being a swarm of local oss models post gpt-5
0
1
9
now that gpt-5 is out i feel like genie 3 was the most impressive release so far this week.
Yesterday we announced Genie 3. One feature of the model that's especially fun to play with is starting worlds from existing videos. Here's a drone shot generated by Veo 3, with me taking control mid-flight.
0
0
4
RT @balajis: Introducing AI agents alongside humans on the Internet is like introducing cars alongside humans on physical streets. (1) Som….
0
193
0
we got lots of work to do to change this narrative about ai agents.
If I hear people talk about "AI agents" these days it's generally a red flag and I know they're non-technical ppl reading AI news but not actually shipping anything. Not cause I don't believe in AI agents but it's such a marketing term with no real meaning at this point.
0
0
4
llm sycophancy needs to be treated during post-training.
Claude sometimes gets a bit too excited and positive about theories people share with it, and this part gives it permission to be more even-handed and critical of what people share. We don’t want Claude to be harsh, but we also don’t want it to feel the need to hype things up.
0
0
0
gpt-oss-20b is an MoE with only 4 experts active per token and 3B parameters active per inference step.
Next to Qwen3 of comparable size: Looks like gpt-oss is a wide (vs deep) model
1
0
6
gpt-oss-20b proposes an ai agent framework architecture 🤯
0
0
4
RT @gneubig: Summary of GPT-OSS architectural innovations:. 1. sliding window attention (ref: .2. mixture of expert….
arxiv.org
Deploying Large Language Models (LLMs) in streaming applications such as multi-round dialogue, where long interactions are expected, is urgently needed but poses two major challenges. Firstly,...
0
360
0