TITweetIndex

Deepak Pathak

@deepakpathak · Pittsburgh, PA · joined 27 May 2013

Co-Founder & CEO @SkildAI, Faculty @CarnegieMellon. PhD @UCBerkeley; BTech @IITKanpur I study topics in AI (robotics, machine learning & computer vision).

29 928Followers
419Following
868Posts total
223.9KViews on collected posts

Against accounts of the same size

8 posts from the last 90 days, next to the 10K–100K follower range. shown widely, but few of those viewers react.

MetricThis accountMedian for 10K–100KRatio
Median views per post16 4722 7436.01×
Reach (views ÷ followers)55.04%8.91%6.18×
Engagement rate0.94%1.64%0.57×

Others in this range →   Compare with another account →   How these benchmarks are built →

Latest posts

One surprising aspect of S1’s in-context learning (ICL) is where it shines most: super long-horizon tasks (10+ min) and scenarios outside the pretraining distribution. For short, simple pick-and-place tasks (5–20 sec), most frontier models can already perform well either https:/ 16.3K views · 155 likes · 30 reposts · 5 replies 29 Aug 2026 Feels nice when folks actually read a long blog carefully. Makes countless hours spent with the team obsessing over every single word totally worth it! This is something I learned the hard way over the years from my dear advisor -- Alyosha Efros. Those who know him, know. Haha. 14.4K views · 138 likes · 6 reposts · 3 replies 27 Aug 2026 We’re scaling S1 as fast as we can. The curve shows no signs of slowing down, steadily going up. We’ve been working for a while to find this kind of scaling law in robotics. 11.1K views · 134 likes · 26 reposts · 6 replies 26 Aug 2026 This is a good question. If the task was only 5–10s long, it can be difficult to distinguish true physical steerability from generalization. But we’re doing tasks that run for over 10 minutes and not seen during training (OOD). Sustaining performance at such long horizon and 12.7K views · 80 likes · 4 reposts · 3 replies 26 Aug 2026 What a crazy statement. Makes me wonder whether Dyna did much data filtering for their model to reach 1M egocentric hours. https://t.co/nq6RTeKXoI 42.9K views · 101 likes · 4 reposts · 6 replies 25 Aug 2026 Very exciting! But I still have the same question for @SkildAI as I did for Generalist's GEN-1.5: does S1 exhibit *physical prompt steerability*, where different prompts induce distinct behavior in the *same* environment? I like Figs. 5 & 6 because they start to get at this 16.6K views · 38 likes · 1 reposts · 1 replies 25 Aug 2026 Scaling laws for in-context learning for robotics, perhaps the most exciting result in robotics https://t.co/jJYxAf9AzX 37.5K views · 397 likes · 35 reposts · 8 replies 25 Aug 2026 In-context learning for robotics is here. - Long-horizon tasks over 10 minutes long - Never seen during pre-training - Prompted with one video, no fine-tuning We are building intelligence from the foundations up, not from the top down. 72.2K views · 514 likes · 54 reposts · 33 replies 25 Aug 2026

Similar accounts