Param Thakkar

Playing with RL, World Models + GPUs, CS undergrad.

[kache @yacineMTB]

wrong. it's simulators

[Param Thakkar @Param23072004]

This is great 😃

[Radiance Fields @RadianceFields]

Apple just announced gaussian splatting is coming to Apple Maps at WWDC.

[Boshen @boshen_c]

One day I will teach my son how persistence will compound.

[Param Thakkar @Param23072004]

🙂🙂

[khushal @khushaltwt]

HackerRank open-sourced their ATS system, and someone built a website around it to score resumes.

I ran my resume through it and got this score 🤔

LMFAO, is this thing even accurate?

Link in the comments 👇

[Ilya Sutskever @ilyasut]

if you value intelligence above all other human qualities, you’re gonna have a bad time

[Daksh Gupta @dakshgup]

boris cherny goes on a podcast every three months and says something like “i’ve stopped breathing now i just wrote a breath.md” and the next day everyone in sf stops breathing

[Param Thakkar @Param23072004]

@reach_vb There still isn't a feature where for example I am working on a PR and I merged it and due to that PR I have merge conflicts in the other PR but codex is neither able to pull the latest changes from main not detect if there are any merge conflicts in order to fix them

[Param Thakkar @Param23072004]

@kernel_trick You just need to apply and they will let you know 🙂

[Nathan Lambert @natolambert]

Something to show people that don't get AI safety at least a little bit. We have so much we don't know and don't currently control in the models.

(extreme content warning, but you're on X)

[Param Thakkar @Param23072004]

Documentation: https://t.co/jXIcIDa6Fm
Release notes: https://t.co/9tO49kZX5L

[Param Thakkar @Param23072004]

TorchWM v0.4.1 is out, this release includes, a bunch of bug fixes found from the previous release, support for DMLab, Brax, Mujoco and Gymnasium Robotics environments, support for torchwm CLI and optimizations for Genie WM.

Check this out: https://t.co/16yiW7EORO

[Param Thakkar @Param23072004]

https://t.co/tQp2pHrcTs

Gamma world, a really nice paper Multi agent world models from Nvidia. They introduced two methods Sparse Hub attention and Simplex Rotary Agent Encoding which lets the world model expand to multi agent scenarios in linear time

[Zian Wang @zianwang97]

🚀 What if physical AI policies could interact with generated worlds in real time?

Introducing OmniDreams, a generative world model for closed-loop autonomous vehicle simulation.

Tech report, code, models, and data samples are available now.

Project: https://t.co/BOTWdSJKMx
Code: https://t.co/hPH3KbE6Uy
Model: https://t.co/G4g9TWFD2W

Join the #omnidreams discord channel: https://t.co/AIwYQvc0bv

[Param Thakkar @Param23072004]

https://t.co/tQp2pHrcTs

Gamma world, a really nice paper Multi agent world models from Nvidia. They introduced two methods Sparse Hub attention and Simplex Rotary Agent Encoding which lets the world model expand to multi agent scenarios in linear time

[abdel @AbdelStark]

Why LLMs are a dead end for human-level intelligence, and especially for Physical AI / Robotics.

The next leap isn’t bigger language models.

It’s World Models.

I just dropped a full 1-hour presentation from Shanghai:
“World Models: the ChatGPT moment for robotics?”

→ Why LLMs hit a wall
→ Why action-conditioned world models planning in latent space are the real path
→ Live World Forge demo with LeWorldModel + Hugging Face LeRobot

Watch here.

The future of intelligence is embodied, not just chatty.

[Param Thakkar @Param23072004]

@danijarh @AiEleuther Links to the session and slides:

Slides: https://t.co/z36sOqWNDQ

Session:
https://t.co/Fy4hXpAxWl

[Param Thakkar @Param23072004]

Delivered a talk on the introduction of world models and DreamerV1 architecture developed by @danijarh in the World Models Reading Group kickoff @AiEleuther .

It was really great getting into the depth of dreamer architecture and introducing world models to people

[mrdoob @mrdoob]

Ported Google's Draco decoder to pure JavaScript.

https://t.co/jbXeB6AzsF

4.3× smaller than the WASM build, byte-for-byte identical output, often faster once you factor in load, init and parse.

[Caglar @caglar_ee]

Video lectures, Umich ROB 311 How to Build Robots and Make Them Move fall 2022, by Elliott Rouse
https://t.co/KrLK8t7ZvS
https://t.co/8reS05Xl5b
.

[Param Thakkar @Param23072004]

@_vmlops Robotics and GPU programming