Nurma

@higgsfield_ai Wow!

@Tim_Qian Cool!

Rewrote bambot.org and the playgrounds using React and Next.js. Finally, I can confidently add more features like camera support, AI control, and more...

@NVIDIAAIDev brick

🎉 Last call ... It's almost time to share your great ideas to #RenameProjectDIGITS with Jensen.👀 🍿We love the creativity.

Do you remember when you joined X? I do! #MyXAnniversary

@pulkitology In 2022 we did exactly the same experiment with humanoid: BC + residual RL youtube.com/watch?v=BMEfo8…

Overcoming the lack of reliability of Behavior cloning (BC) with reactive reinforcement learning. Action-chunking is a two-edged sword- it's critical for BC to work, but it also limits how adaptive the robot is to disturbances and corner cases. Learn more: residual-assembly.github.io Work led by @larsankile along with @anthonysimeono_ @IdanShenfeld and @marceltornev

@higgsfield_ai @nebiusofficial Nice!

@xf1280 I'm fascinated by your research. Would it be possible for you to share the simulation environment for dexterous manipulator robot? I believe it would significantly benefit my research.

🤖Excited to share our project where we propose to use rewards represented in code as a flexible interface between LLMs and an optimization-based motion controller. website: language-to-reward.github.io Want to learn more about how we make a robot dog do moonwalk MJ style?🕺🕺

@_akhaliq Mind blowing

Symmetry-driven 3D Reconstruction from Concept Sketches paper: repo-sam.inria.fr/d3/SymmetrySke… project page: ns.inria.fr/d3/SymmetrySke… gitlab: gitlab.inria.fr/fhahnlei/symme…

@robo_challenge I prepared some solutions, but couldn't find how to submit it. Can you please help with this?

[3/3] Real Robot Stage: you get access to our cluster of robots to test your algorithms. The data now comes from real robots. We have data from a well-performing policy and from mixed performance. #OfflineRL #Robot #MachineLearning #DeepLearning

It is time to put offline reinforcement learning methods to the test in the real world! Take part in our real-robot-challenge.com @NeurIPS competition starting today. Your task is to use pre-recorded data to learn how to manipulate a cube with the TriFinger robot @MPI_IS. [1/3]🧵

@facebookai Does it work with batch of images?

We’ve just open-sourced AugLy, a new #Python library that will help AI researchers use data augmentations to evaluate and improve the robustness of their machine learning models. Read more: ai.facebook.com/blog/augly-a-n…

Вы помните, когда зарегистрировались в Твиттере? Я помню! #МояГодовщинаВТвиттере

@ARGleave @iclr_conf Until now I did not believe in contactless fight. But now you have changed my mind😂 Some funny fights youtube.com/watch?v=BFUqgb…

Excited to share our work @iclr_conf: deep RL policies can be attacked by another agent taking actions that create natural adversarial observations. (1/5) Blog: bair.berkeley.edu/blog/2020/03/2… Paper: openreview.net/forum?id=HJgEM… Videos: adversarialpolicies.github.io Code: github.com/humancompatibl…

Deep RL agents are data hungry and often learn task-specific representations. Our model learns object-centric abstractions from raw videos. This enables highly data-efficient RL and structured exploration. arxiv.org/abs/1906.11883

.@Kaplanyan has been killing it with these foveated reconstruction papers. Absolutely amazing that this is possible with so little data. Have a look! - ai.facebook.com/blog/deepfovea…

Dataset: 64 minutes of challenging pedestrian areas captured with: - stereo cylindrical 360 RGB cam - 3D point clouds from two Veloydyne 16 Lidars - line 3Dpoint clouds from two Sick Lidars - Audio, RGBD, 360 spherical image from a fisheye camera jrdb.stanford.edu/dataset/about

We’ve developed one of the first systems that uses only 3D point clouds and achieves higher precision in detecting 3D objects than prior work. Read more: ai.facebook.com/blog/pushing-s…. And we’ve open-sourced this model here: github.com/facebookresear… #ICCV2019

Celeb-DF: A New Dataset for DeepFake Forensics pdf: arxiv.org/pdf/1909.12962… abs: arxiv.org/abs/1909.12962 github: github.com/danmohaha/cele…

It has been in our TODO stack for an eternity now… So excited that we are taking the time to write a paper for our 🤗Transformers library! Stay tuned, you’ll finally have a citable paper very soon! 📃

miniBERT, a multi lingual BERT from @GoogleAI , 27X times more rapid on CPU. Same approach (distillation of BERT LM) than few days old DistilledBERT from @huggingface and @spacy_io plugin. Got SOTA on 2 seq tagging tasks. Obviously there is a trend :-) arxiv.org/pdf/1909.00100…