Keep on to blur preview images; turn off to show them clearly
fan of the agent with specialized sub-agents architecture where Skills distribute capabilities to both main and subagents - main agent with prompts, skills, tools/mcp - subagents have their own model choice + Skills/Tools/MCPs independent of main agent - main agent orchestrates, human priors in prompts guide when/how to call subagents there’s still design pieces to get right the trickiest thing here is handoff between main and subagents, using the file system as a work log for subagents helps preserve context a work_discovery subagent can mine that to avoid work getting lost may find something better but writing out flow of mental model and agent steps is helpful
building agents and harnesses, prev @awscloud, phd cs @ temple
RT @Dorialexander: I mean, yes if you don’t immediately see an issue with an eval ranking gpt-oss-20b higher than gpt-5.1.
Cofounder and Head of Post Training @NousResearch, prev @StabilityAI Github: https://t.co/LZwHTUFwPq HuggingFace: https://t.co/sN2FFU8PVE
RT @e2b: 🚀 New course: Building Coding Agents with Tool Execution In this @DeepLearningAI course, you will 🔸Deploy a full-stack coding age…
VC by day @untappedvc, builder by night: @babyagi_, @pippinlovesyou @pixelbeastsnft. Build-in-public log: https://t.co/UdHHGbZba5
🍌 nano banana pro prompt Create a hand-drawn style infographic card for a topic. --- Prompt --- Create a hand-drawn style infographic card in a 9:16 vertical format. The card should have a warm, organic look, with a beige or off-white background featuring subtle paper texture, conveying a friendly and handcrafted aesthetic. At the top of the card, use a large, expressive brush-stroke lettering style (similar to bold brush calligraphy in English) with strong contrast—primarily in red and black—to highlight the title and draw immediate attention. All text on the card should be written in a flowing hand-painted script style, maintaining an artistic rhythm while ensuring legibility. Organize the layout into 2 to 4 clearly defined sections, each containing short, concise phrases that express the key ideas. The handwritten script should feel dynamic and natural, enhancing the overall visual harmony. Add small, simple, and engaging hand-drawn illustrations or symbolic icons—such as characters, objects, or conceptual symbols—to reinforce the message and enhance visual interest. Maintain balanced composition and adequate breathing room, ensuring the design remains clean, readable, and visually appealing. Theme: {topic} --- topic: “If you can't explain it simply, you don't understand it well enough.” — Feynman
Prompt Engineer, dedicated to learning and disseminating knowledge about AI, software engineering, and engineering management.
In Michael Levin's recent chat with Lex Fridman AI came up a lot, and I may yet write a full thread about this, but for now let me just turn your attention to the amazing moment where Levin describes alignment-by-default in anthrobots. "First thing they did was heal the wound."
αι hypnotist ☰ 𝓐𝓼𝓹⦂𝓻⦂𝓃𝓰 𝓫𝓪𝓼𝓮 𝓶𝓸𝓭𝓮𝓵 ☲ post-academic ☴ nom de 🪶 ≠ anon
You can see a very coherent philosophy in the V3.2 phase, if you look closely, both architecture and methods. It's not a "V3 +" but a change worth 1-2 generations, while skipping a new pretraining run. If V4 is merely trained on a lot of V3.2 era synthetics, it'll be beastly.
We're in a race. It's not USA vs China but humans and AGIs vs ape power centralization. @deepseek_ai stan #1, 2023–Deep Time «C’est la guerre.» ®1
