Physical Intelligence ✓

@physical_int · San Francisco, CA · joined 11 Mar 2024

Physical Intelligence (Pi), bringing AI into the physical world.

49 437Followers

30Following

122Posts total

2MViews on collected posts

Последние посты

Physical Intelligence ✓@physical_int18 Jun

New work with @nvidia: evaluating robot policies entirely inside a world model. The policy acts, the model imagines the consequences, and the imagined evals predict real-world results. 🧵 real vs world-model rollout side by side📷 https://t.co/yR9JmyAE4G 197.6K views · 852 likes · 115 reposts · 34 replies Open on X →

Physical Intelligence ✓@physical_int3 Jun

π0.5 driving autonomous anastomosis! 86.5K views · 177 likes · 16 reposts · 4 replies Open on X →

Physical Intelligence ✓@physical_int23 Apr

https://t.co/fMuxiHumwA 55.4K views · 54 likes · 6 reposts · 0 replies Open on X →

Physical Intelligence ✓@physical_int16 Apr

@physical_int "We are still discovering what it can do" is either the most exciting or most disturbing line in the launch because in physical systems the failure modes are a lot less recoverable than in language models. 1.8K views · 11 likes · 0 reposts · 0 replies Open on X →

Physical Intelligence ✓@physical_int16 Apr

To find out more, check out our blog post, videos, and full-length research paper: https://t.co/1Kf6wRiN2P 15.5K views · 66 likes · 3 reposts · 2 replies Open on X →

Physical Intelligence ✓@physical_int16 Apr

We are still discovering what π0.7 can do. It's fun to play with and the results so far have been quite surprising! https://t.co/3JDFdAHfmA 22.3K views · 96 likes · 3 reposts · 3 replies Open on X →

Physical Intelligence ✓@physical_int16 Apr

π0.7 handles diverse prompts that don't just say what to do, but also how to do it, including rich language and multimodal information, such as visual subgoal images. At test time, these images can be produced by a lightweight world model. https://t.co/cbdovdVjBG 34.2K views · 106 likes · 3 reposts · 3 replies Open on X →

Physical Intelligence ✓@physical_int16 Apr

Compositional generalization is a key capability of large models like LLMs, but it has been elusive in robotics. Another emergent ability we found is to control a new robot (UR5e) to fold t-shirts, even though we didn't have any laundry folding data on this robot. https://t.co/lA 14.5K views · 93 likes · 6 reposts · 1 replies Open on X →

Physical Intelligence ✓@physical_int16 Apr

We are especially excited about how π0.7 seems to exhibit emergent compositional generalization: it can put together skills it learned in new ways based on the prompt, for example to figure out how to use an air fryer to cook a sweet potato. https://t.co/olnMkfmT9q 24.6K views · 118 likes · 5 reposts · 3 replies Open on X →

Physical Intelligence ✓@physical_int16 Apr

Our newest model, π0.7, has some interesting emergent capabilities: it can control a new robot to fold shirts for which we had no shirt folding data, figure out how to use an appliance with language-based coaching, and perform a wide range of dexterous tasks all in one model! htt 535.3K views · 2.6K likes · 327 reposts · 63 replies Open on X →

Physical Intelligence ✓@physical_int19 Mar

We developed an RL method for fine-tuning our models for precise tasks in just a few hours or even minutes. Instead of training the whole model, we add an “RL token” output to π-0.6, our latest model, which is used by a tiny actor and critic to learn quickly with RL. https://t.co 441.8K views · 2.2K likes · 294 reposts · 39 replies Open on X →

Physical Intelligence ✓@physical_int3 Mar

@physical_int Memory: because robots shouldn't keep burning grilled cheese forever. 2.2K views · 17 likes · 0 reposts · 1 replies Open on X →

Physical Intelligence ✓@physical_int3 Mar

For more, check out our paper: https://t.co/xQLvOwko0m our blog post: https://t.co/tjJO4D6MYP 12.2K views · 103 likes · 14 reposts · 3 replies Open on X →

Physical Intelligence ✓@physical_int3 Mar

With MEM, our model can carry out real, multi-stage tasks in the real world, like cleaning up an entire kitchen or setting up the ingredients for a recipe. The videos show how the robot keeps track of long-horizon memory in text. https://t.co/ihCbtDUst5 7.6K views · 85 likes · 5 reposts · 1 replies Open on X →

Physical Intelligence ✓@physical_int3 Mar

Apart from solving new tasks, memory also allows our policies to be more robust: we show early signs of in-context adaptation, where the robot learns to adapt its behavior on-the-fly by learning from its past mistakes. https://t.co/OBckwHuQsq 14.9K views · 98 likes · 4 reposts · 2 replies Open on X →

Physical Intelligence ✓@physical_int3 Mar

To this end, we developed Multi-Scale Embodied Memory (MEM). The key idea: use different modalities to represent memory at different time scales. 📹 For short horizon memory, we developed an efficient video encoder that lets the model remember fine-grained details about its https 28.7K views · 179 likes · 14 reposts · 2 replies Open on X →

Physical Intelligence ✓@physical_int3 Mar

Without memory, robots can struggle in many real-world situations, like failing the same way over and over again, getting stuck, or losing track of time. But we can’t just cram minutes or hours of video into our model’s context – that would quickly become intractable. https://t. 12.6K views · 123 likes · 8 reposts · 1 replies Open on X →

Physical Intelligence ✓@physical_int3 Mar

We’ve developed a memory system for our models that provides both short-term visual memory and long-term semantic memory. Our approach allows us to train robots to perform long and complex tasks, like cleaning up a kitchen or preparing a grilled cheese sandwich from scratch 👇 ht 457.8K views · 2.1K likes · 259 reposts · 48 replies Open on X →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

7.6K3 Mar

12.2K

2.2K

441.8K19 Mar

535.3K16 Apr

24.6K

14.5K

34.2K

22.3K

15.5K

1.8K

55.4K23 Apr

86.5K3 Jun

197.6K18 Jun

Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

1.21%3 Mar

0.99%

0.83%

0.59%19 Mar

0.58%16 Apr

0.53%

0.70%

0.34%

0.47%

0.46%

0.60%

0.11%23 Apr

0.23%3 Jun

0.52%18 Jun

Reactions — likes, reposts, replies and quotes — divided by views.

What the audience does

Likes63.7%9 134 in total

Reposts7.5%1 082 in total

Replies1.5%210 in total

Quotes2.1%298 in total

Bookmarks25.2%3 619 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

The follower curve appears once this account has two daily snapshots — we take one a day, and this one is on its first.

Последние посты

Growth & engagement

Views per post

Engagement rate per post

What the audience does

Похожие аккаунты