Andement (@Andee_Park) header
Andement (@Andee_Park) profile picture

physics and nature @naturewill

Joined November 2021
Photo by Andement (@Andee_Park)Photo by Andement (@Andee_Park)Photo by Andement (@Andee_Park)Photo by Andement (@Andee_Park)Photo by Andement (@Andee_Park)Photo by Andement (@Andee_Park)Photo by Andement (@Andee_Park)Photo by Andement (@Andee_Park)Photo by Andement (@Andee_Park)Photo by Andement (@Andee_Park)
A thread on how we built the infrastructure behind Natura-1.
1
3
124
5/ The inference side is where it gets interesting for a physical robot. A Rust orchestrator layer overlaps observation streaming, computation, and action transfer simultaneously. Fused CUDA kernels cut latency on the critical path. The result: a strict 100ms observation-to-action loop on physical hardware. That gap is the difference between a movement that feels natural and one that feels mechanical.
1
31
The part worth emphasizing: every component was designed to scale further. Natura-1 is among the first models trained on this stack — not the last.
20
We just published the technical blog for Natura-1 — our first general foundation model for the physical world. Here's what it actually means, for anyone who doesn't speak robot-researcher. 🧵
2
6
295
7/ The part I find most interesting: Most foundation models learn from text. Natura-1 learns from what humans do — a million hours of hands at work. That's a fundamentally different bet on what intelligence actually is.
1
18
It is actually a great great experience to be the first person to play with Natura-1 long-shot. Learning how a model/robot acts is like learning how to chat with you agent, you are gonna find the best ways to communicate with each other. Natura-1, my first intelligent friend.
30 minutes. One take. No cuts. We put Natura-1’s zero-shot generalization to the test with complex spoken instructions and unfamiliar tasks and environments. It responds to new requests, adjusts and retries when an action fails, and keeps going autonomously.
2
14
981
We are excited to launch Natura-1: a 50B model pretrained on 1M hours of human experience, turning language into dexterous action, zero-shot. So proud to be able to bring this launch to life. What Nature Will has built is incredible. So excited to share it with the world.
Introducing Natura-1, our first general foundation model for the physical world. It follows natural-language instructions zero-shot and turns them into action with dexterous hands. Natura-1 is a 50-billion-parameter mixture-of-experts (MoE) model pretrained on one million hours of egocentric human experience. On top of this foundation, we conduct efficient post-training on a limited amount of robot data to adapt pretrained world knowledge to dexterous manipulation, yielding a model that generalizes to a wide range of unseen scenarios. Natura-1 follows complex natural-language instructions zero-shot, reasoning through spatial constraints, multi-object relationships, and negative conditions. It can also autonomously carry out long and continuous workflows to perform long-horizon executions. Natura-1 provides early evidence that generalization—and even emergent behavior—can coexist with the complexity of dexterous hands. Further progress depends on the full stack: infrastructure for faster training and real-time inference, pretraining data at greater scale and coverage, and models that transfer to robots with less adaptation. We expect it to be a long-term research and engineering effort.
1
13
1,212
nature will perceive…
nature will _______.
4
236
The Definitive Guide to Harness Engineering. "We are shifting from being the creators of code to becoming the guardians of the creation process." This should be our stance in this new era.
3
700
My man @pxeight 's business is now officially live here. Marketers and builders, this is the guy you should work with, a man who shaped how @PythNetwork looks today. Pythians are always the best 🔮
PX8 Collective is live. The focus is simple: graphics and motion. Now open for new projects.
1
6
909
One thing that makes me very bullish on Chinese builders is that - they never say no to anything. You’ll hear them saying “we’ll find a way” instead of “I cannot do this”. You may say it’s risky, not regulated, or opportunistic. But many things work out actually in this way.
5
398